1. Introduction
Pneumoperitoneum (PNP) is a common clinical entity defined by the presence of free air or gas in the peritoneal cavity [
1]. Gastrointestinal (GI) tract perforations represent an important cause of PNP, a potentially life-threatening condition associated with high morbidity and mortality (30–50%), with even poorer outcomes when the diagnosis is delayed [
2,
3]. As documented in the previous literature, GI tract perforations can have various causes, including trauma and iatrogenic injury, inflammatory conditions, infection, ischemic change, diverticula, foreign bodies and malignancy [
4,
5]. It is important to correctly identify the site and the cause of the perforation in order to appropriately manage and make decisions regarding surgical planning. However, clinical diagnosis of the GI perforation site may be difficult, as symptoms may be non-specific (abdominal pain and distension, vomiting, constipation, fever, diarrhea, tachycardia, hypotension, and tachypnea) and related to multiple factors, including the source of the perforation and its mechanism, time elapsed since the perforation, the degree of contamination of the peritoneal cavity and the patient’s age [
1]. In the emergency setting, radiology is vital in the early detection of this pathologic condition [
6].
Until now, the abdominal upright posteroanterior X-ray is traditionally regarded as the foremost method of detection, with the relevant signs described in [
7,
8].
However, extraluminal air is not always demonstrable on plain abdominal radiography, especially if self-sealed, well contained by adjacent organs, or if the perforation is too early or small, considering that sensitivity collapses (50–70%) for GI perforation with less than 1 mL of gas [
9]. The patient’s clinical status could also make the recognition of free air difficult, reducing the method’s diagnostic value in patients too sick or debilitated to stand up for an erect abdominal plain film [
8]. Only 55–85% of PNP could be detected by abdominal X-ray plain film [
8].
Computed tomography (CT) is the modality of choice in the detection of GI tract perforation, due to its high sensitivity and accuracy (82 to 90%); it can display intra- and extra-peritoneal air, localizing the perforation site [
10]. CT demonstration of PNP occurs when free gas, discontinuity of the GI wall or leakage of orally administered contrast medium are visible [
11]. Other signs may include fluid abdominal effusion [
12], inhomogeneous mesentery, wall thickening, “dirty mass” (extra-luminal fecal matter) [
13], intestinal or porto-mesenteric pneumatosis, and abdominal abscess [
8,
14]. A total of 83–100% of PNP could be diagnosed through CT [
6], although it is not cost effective and is associated with radiation exposure.
Ultrasound (US) findings of PNP were first identified in 1984 [
15]. It has been profusely demonstrated that US could be useful in identifying PNP and investigating its causes [
6,
9], as reported in several studies and case reports [
16,
17,
18], attesting to very different values of diagnostic accuracy (53–100%) [
19,
20].
In free abdominal air detection with US, the use of the linear array transducers (10–12 MHz) represents the most sensitive standard of detection, thanks to its high resolution [
21]. The supine position with the thorax slightly elevated (10–20 degrees), and the prone position exploring the right paramedian epigastric area, are the best methods, while the right upper quadrant and the pre-hepatic space are the most common sites of air accumulation [
22].
The US detects signs of free abdominal air via the scattering of US waves in correspondence with the anatomic interfaces between soft tissue and bubble gas, associated with reverberation phenomena of the waves between the transducer and air bubbles. This event produces an increased echogenicity of a peritoneal stripe accompanied by numerous reflection artefacts, typically with a ring-down or comet-tail appearance, associated with the characteristic feature of real-time modifications as the patient’s position changes [
6,
21,
22,
23].
The diagnostic confidence in US rises when increased thickness of the bowel wall, collected fluid effusion and a reduction in peristaltic moves are also present [
24]; on the other hand, intraluminal air may be recognizable when peristalsis and normal wall thickness are present. The detection of peritoneal stripes, especially if modified by the patient’s position change, is very suspect for PNP [
25]. In addition, US permits one to observe the motion of free air in real time, to discriminate the air in the lungs from PNP, thanks to respiration excursion [
22], to localize the presence of retroperitoneal perforation detected by air around the duodenum and the pancreatic head [
26], and to use specific US methods, such as the scissor maneuver by Karahan [
27].
Moreover, US is widely recognized as an indispensable tool in the bedside diagnosis of the acute abdomen, and in the trauma context (FAST), where it has been recommended as standard procedure [
28]. It may also be useful in patients where radiation burden should be limited, for example, children and pregnant women [
6]. When compared to plain radiography, US demonstrates greater sensitivity (93% vs. 79%, respectively), very similar specificity (64%), and a positive predictive value (97%) [
29].
Despite this, many text books and lecturers, especially in the field of emergency and critical care, skip the topic entirely. US has not been adequately integrated into the standard diagnostic process of PNP detection. The usual explanation for this is that gas is a strong reflector, able to prevent the transmission of US waves and create reverberation artifacts, inhibiting the obtainment of some diagnostic information. In addition, the physiological gas within the bowel may make it even more difficult to obtain an accurate interpretation [
24].
The aim of this study was to evaluate the usefulness of a new diagnostic US tool in PNP detection, based on the imaging performed by the US contrast-specific software, which is the software generally used during contrast-enhanced US examination (CEUS). In our series of tests, the contrast-specific software was routinely used in the emergency setting as the first tool to assess any patient with an acute abdomen. This was performed in an innovative way, as it was not dependent on the administration of contrast agents, and was defined as contrast-specific mode (C-mode). The C-mode operates in real-time, with a low mechanical index (pulse inversion technology), and uses a digital subtraction in order to isolate the signal in double harmonic. [
30] Thanks to this method, the receiver electronically filters and deletes the fundamental frequency, showing only the double harmonic signal on the monitor, which generally originates from the vessels and from the air bubbles in the abdominal cavity in our specific case. Our intent is to help the radiologist diagnose abdominal free air with confidence, implementing US B-mode evaluation through the use of C-mode, which strongly emphasizes the sonographic air signal.
2. Methods
Our study included 157 consecutive patients who arrived in our emergency department with acute abdomen between April 2018 and October 2019. Hemodynamic instability and renal failure were considered exclusion criteria. All the included patients underwent US examination first, divided into two steps, and then a CT examination (
Figure 1). This workflow did not cause any significant time delay in the diagnosis and management decision process.
The US examination was performed using a Resona 7 system (Shenzhen Mindray Bio-Medical Electronic Co, Shenzhen, China), equipped with both curved- and linear-array probes, and with dedicated software for acquisition of contrast-specific imaging. No administration of US contrast media was used.
All CT examinations were performed with a 128-slice DSCT (Somatom Definition Flash; Siemens Healthcare, Forchheim, Germany) after iodinated contrast media administration (Iomeprol injectable solution, Iomeron 400); images were acquired in a single portal venous phase.
The US examination was performed by two different radiologists with no less than 5 years of experience in the emergency department (R1 and R2); CT evaluation was performed by a radiologist with 25 years of experience (R3).
The US examination was divided into two steps, consisting, first, of an examination performed using baseline US (B-mode), and a second examination using the mentioned US technology consisting of C-mode, with no contrast media administration, to focus on the free air detection. R1 and R2 conducted the two steps of the US examination for each patient; the interobserver concordance rate was calculated to explore the diagnostic performance and reproducibility.
All patients underwent CT examination after the US study was performed by R1 and R2 and the CT was examined by R3. Radiologists were blinded to the results of the other examinations. Since CT is considered as the gold standard in free air detection, the statistical analysis allowed us to estimate the sensitivity, specificity and diagnostic accuracy for CT, US B-mode and C-mode. Positive and negative predictive values of US B-mode and C-mode were also evaluated. Written informed consent was obtained from each patient.
3. Results
Our population of study included 157 patients with an acute abdomen (96 M and 61 F; mean age 41 ± 16 years). The main characteristics are described in
Table 1.
In this study, 32 out of 157 patients (20%) received surgical confirmation of GI perforation. The other causes of an acute abdomen were acute diverticulitis (n = 39), intestinal occlusion (n = 24), acute appendicitis (n = 18), acute pancreatitis (n = 14), acute cholecystitis (n = 12), ischemic colitis (n = 5), renal colic (n = 5), intestinal volvulus (n = 3), inflammatory bowel disease (n = 3), and ovarian torsion (n = 2).
A statistical analysis of the results was conducted in order to calculate the sensitivity, specificity, diagnostic accuracy, and positive and negative predictive value of the different methods. Therefore, the concordance index using the Choen’s k test for C-mode and B-mode in the detection of free air was calculated.
In our population, CT correctly detected 31 out of 32 patients with GI perforation.
The CT value of sensitivity and specificity in free air detection were 97% and 100%, respectively, with a high level of diagnostic accuracy (99%) in showing free air.
In the analysis results of both R1 and R2, C-mode US demonstrated higher average values than B-mode US in sensitivity (93% vs. 70%, respectively), specificity (98% vs. 88%, respectively) and diagnostic accuracy (97% vs. 81%, respectively). C-mode US identified 30 perforated patients in R1’s examinations and 29 in R2’s examinations, while B-mode reached the diagnosis in 23 and 21 cases, respectively, in R1 and R2’s examinations.
B-mode US examinations performed by R1 erroneously suspected the presence of PNP in 9 cases (6% false positive), while R2 obtained 10 false positives (6.4%). C-mode US obtained only two false positives (1.3%) from R1 and three false positives (1.9%) from R2.
In no case did any CTs performed by R3 result in a false positive.
On the other hand, the false negative average value for B-mode and C-mode was, respectively, around 6% and 1.6%, while the false negative rate for CT examinations was 0.6% (1 case). Therefore, the negative predictive average value was 98% for C-mode US and 93% for B-mode US, while the positive predictive value was 94% for C-mode US and 71% for B-mode US.
The interobserver concordance index for C-mode between R1 and R2 was 99% (Cohen’s k: 0.979); the same index for B-mode was 99% (Cohen’s k: 0.946). The R1 intraobserver concordance index for C-mode and B-mode detection of PNP was 77%, and the R2 intraobserver concordance index for C-mode and B-mode detection of PNP was 74%, but the Choen’s k results were unsatisfactory in both (0.175 and 0.253, respectively), confirming the limited validity of B-mode US as a diagnostic test for air detection.
Concerning the localization of free air bubbles, in our cases, the preferred sites to exalt the presence of free abdominal air were found to be the interface with the anterior abdominal wall, hepatic interspace (peri-hepatic profiles, falciform ligament and hepatic hilum cavity), and peri-duodenal space.
4. Discussion
The rationale of our study was to utilize the strengths of US in the assessment of PNP, and demonstrate its diagnostic power without the use of contrast media; to date, we have not found similar studies in the literature.
In most European emergency departments, US is routinely performed by radiologists in their first assessment of any patient with an acute abdomen [
31].
Today, the majority of modern US machines are already equipped with specific software for CEUS, with a large possibility of scan acquisition and co-registration after the administration of contrast agents.
The evaluation of B-mode plus C-mode examination was important to understand if any significant diagnostic information was added, as confirmed in our results.
One patient, whose micro-perforated duodenal ulcer was found via surgery, could not be diagnosed by CT, probably due to the low amounts of sufficiently detectable free gas and the poor communication of indirect signs. In the same patient, C-mode and B-mode were also not able to provide the diagnosis for both US examinators.
As reported in our results, 30 patients were correctly detected on C-mode by R1. In two patients, C-mode was not able to reach the diagnosis of GI perforation; surgical intervention confirmed the presence of a micro-perforated duodenal ulcer and self-sealed peri-diverticular air, respectively. The C-mode performed by R2 identified 29 perforated patients, adding one more missed diagnosis of PNP in a case of peri-duodenum perforation in post-attinic parietal change due to pancreatic carcinoma history.
B-mode US was not able to correctly detect the presence of free abdominal air in 9 and 10 patients, respectively, missing the diagnosis.
Two cases of C-mode false positives were obtained from R1 and R2 examinations, in the first hypothesis, due to intraluminal bowel gas artefacts.
In our experience, the statistical results showed higher sensitivity and specificity in C-mode than in B-mode US (93% vs. 70% and 98% vs. 88%, respectively); taking CT into consideration as the gold standard, with a reported diagnostic accuracy rate of 99%, the results also confirm the high diagnostic accuracy of this innovative tool (97%) and its high negative predictive value (98%).
The high value of the inter-observer concordance index (99%) reflects well upon the reliability of the C-mode method, demonstrating its reproducibility.
The qualitative superiority of C-mode over B-mode is adequately demonstrated by the imaging.
In our patients, larger air bubbles appeared as bright, highly echogenic lines with distal reverberation and shadowing artifacts, as ring-down or comet-tail artifacts on B-mode
(
Figure 2a); free air can also be detected beneath the anterior abdominal wall, where it generally accumulates in the supine patient (
Figure 2b).
C-mode gave a more intense appearance to the peritoneal stripe when compared to B-mode, identifying the amount of free air, similar to the strongly enhanced peritoneal lines in the pre-hepatic space (
Figure 3), and better enounced the presence of intensely enhanced small amounts of air around the falciform ligament (
Figure 4a), as confirmed by the CT scan (
Figure 4b).
Therefore, in some cases of our study population, a small amount of air under the abdominal wall was not clearly visible, and was potentially lost when the examination was not performed in expert hands, where it was misunderstood as a simple anatomic interface on B-mode (
Figure 5a,b). In these cases, C-mode demonstrated its ability to distinctly reveal highlighted peritoneal stripes (
Figure 5c) that modified their aspects as the patient’s position changed (
Figure 5d).
In another case, C-mode was able to detect free abdominal air around the duodenum and the pancreatic head with certainty (
Figure 6a), which was also displayed perfectly by the CT scan (
Figure 6b).
As observed in
Figure 7, an accurate B-mode examination could also lose a minimal amount of free gas; the same patient was correctly detected on C-mode, discriminating a few air bubbles as bright punctuate foci of the luminescent echo line.
Moreover, in one obese patient, C-mode better revealed the presence of PNP and demonstrated its efficient performance, despite the thick fat plane in the anterior abdominal wall and the interposed meteorism, when compared with B-mode.
In addition, in one single case of a child with appendicitis complicated by perforation (excluded from the present study for the lack of CT examination), our diagnosis was confident, thanks to the clear integration of C-mode scans during the US examination; the diagnosis was then surgically confirmed.
As mentioned in our results, air bubbles were most frequently located in the anterior abdominal wall, hepatic interspace (peri-hepatic profiles, falciform ligament and hepatic hilum cavity), and peri-duodenal space. However, the presence of suvra-mesocolic free air does not exclude a sub-mesocolic site of leakage in the GI system. Moreover, one must consider the spontaneous trend of air to move upwards, according to the anti-gravity effect; CT examination results were mandatory in order to detect the leakage.
Although the localization of detected free air bubbles cannot be considered a reliable criteria of C-mode, to obtain deep information about the site of perforation, the benefits of C-mode are proved when compared to B-mode US, as shown in our study. GI perforation does not display specific symptoms, and abdominal X-rays only detect free air in a limited percentage of cases, dependent on the size and progress of the perforation; nevertheless, not all the patients with acute abdomen immediately undergo CT examination. Therefore, any patient who arrives at the E.D. with acute abdomen should receive a C-mode examination, so that PNP may be detected and diagnosed as soon as possible. Consequently, the decision to quickly and decisively utilize CT examination and treatment will save time and reduce the number of people with undiagnosed PNP.
5. Limitations
The limitations of our study are as follows: (1) The small population size of our test group. It will be necessary to apply the same method to a larger population to confirm our results; (3) False negatives. The number of false negatives is small in this study, but represents a limit to exceed in order to increase the reliability of C-mode’s diagnosis of PNP.
6. Conclusions
The use of C-mode US examination in the emergency department demonstrated its superior image quality, and its higher levels of sensitivity, specificity and diagnostic accuracy.
This method did not change the number of necessary scans and the timing of US examination. Furthermore, it has demonstrated that it can reliably diagnose PNP without the use of iodinated radiation or the administration of contrast agents.
In addition, this method’s performance has proven reliable, even when dealing with patients who would otherwise be difficult to diagnose, such as children and the obese. It may, therefore, be used instead of a CT examination, which may be held in reserve should abdominal pain remain undiagnosed after US examination.
In our experience, this new method could provide better visibility of the diagnostic elements of free abdominal air detection, resulting in increased confidence in the diagnosis of PNP; the C-mode US could help young radiologists, or radiologists with poor experience, in emergency clinical situations, to not miss this important diagnosis.