Exploring the next frontier in diagnosing spontaneous bacterial peritonitis

Abstract

Spontaneous bacterial peritonitis (SBP) is a common complication of liver failure. It is an acute bacterial infection of the ascitic fluid in patients with liver cirrhosis. SBP presents a significant challenge for hepatologists owing to its associated complications. While diagnostic paracentesis with polymorphonuclear count is highly accurate, it can be troublesome for some patients as it is an invasive procedure with associated risks. Several studies have proposed new diagnostic methods to improve current practices, many of which remain invasive. Although some serum tests show promise in the diagnosis of SBP, the results are still preliminary. Recent advancements in artificial intelligence and machine learning have introduced predictive models and scoring systems for diagnosis. However, these models still lack sufficient sensitivity, specificity, and the ability to effectively assess treatment response.

Key Words: Spontaneous bacterial peritonitis; Diagnosis; Liver cirrhosis; Ascites; Paracentesis; Procalcitonin; Calprotectin; Liver cell failure; Ascitic fluid

Core Tip: In this editorial, we respond to the retrospective cohort study recently published in the World Journal of Hepatology. The authors highlighted the sensitivity of the diagnosis of spontaneous bacterial peritonitis (SBP) using the automated cell count method. Here, we discuss the advances in SBP diagnosis and the future directions in this field.

INTRODUCTION

Spontaneous bacterial peritonitis (SBP) is a common sequela of liver cell failure. It is considered an acute bacterial infection of the ascitic fluid in patients with liver cirrhosis[1]. Nearly 25% of patients with post-cirrhotic ascites will develop SBP. In approximately 75% of cases, the causative organism is a gram-negative bacterium, with Klebsiella pneumoniae being the predominant organism, accounting for nearly 50% of SBP cases[2].

Patients with liver cirrhosis and ascites should be evaluated for SBP when they present with any acute condition. While the classic symptoms may include abdominal pain, fever, or diarrhea, it is important to note that up to one-third of patients may present asymptomatically[3]. Abdominal pain varies from mild to severe and may be accompanied by signs of peritonitis (guarding, rigidity, and rebound tenderness)[4].

Although first described in the early 20th century, the diagnosis of SBP still relies on invasive procedures, which carry potential risks[5]. SBP is diagnosed based on an ascitic fluid polymorphonuclear leukocyte count of more than 250 cells per mm³. This alone is sufficient to initiate empirical antibiotic treatment. However, a definitive diagnosis also requires a positive ascitic fluid culture and the exclusion of secondary causes of bacterial peritonitis[6].

In this editorial, we respond to the retrospective cohort study recently published in the World Journal of Hepatology. The authors highlighted the sensitivity of the diagnosis of SBP using the automated cell count method[7]. Here, we will discuss the advances in SBP diagnosis and the future directions in this field.

DIAGNOSIS OF SBP

Current diagnostic methods for SBP and their limitations

The diagnosis of SBP currently relies on the analysis of the ascitic fluid obtained through paracentesis. This includes analysis of the polymorphonuclear neutrophils (PMN) count, which has a sensitivity of 93%, and cultures of the ascitic fluid, which could be positive in 40% of cases[8]. It is established that a PMN count of more than 250 cells per mm³ is enough to start empirical antibiotics to treat SBP. However, antibiotics should not be delayed in symptomatic patients or those with a total white cell count of more than 500 cells per mm³, even if the PMN count is below this threshold[9].

The diagnosis of SBP remains challenging owing to the invasive nature of the procedure, which could lead to a delay in diagnosis or misdiagnosis, especially if antibiotics are administered before obtaining a sample from the ascitic fluid[10].

Also, nearly 30% of SBP cases are asymptomatic, making it difficult to perform paracentesis in such patients[11]. Another challenge is that the European Association for the Study of the Liver recommends at least one follow-up diagnostic paracentesis at least 48 hours after starting antibiotics. Although paracentesis is generally considered safe in patients with a high international normalized ratio, its invasive nature still poses risks[12].

Diagnostic sampling should be done properly to avoid contamination of the culture with skin flora and to limit the risk of diagnosing culture-negative neutrocytic ascites[13].

Furthermore, PMN counting is usually done manually, which is time-consuming and requires a certain level of expertise, potentially leading to diagnostic delays[14]. This challenge could be overcome by using an automated cell count method, such as leukocyte esterase reagent strips, which still requires ascitic fluid samples. In a recent meta-analysis[15], the authors reported that the results of these reagent strips are highly comparable to those obtained using the manual method, although differences exist between manufacturers.

NEWLY INTRODUCED METHODS FOR DIAGNOSIS IN ASCITIC FLUID (INVASIVE METHODS)

Recently, new diagnostic methods have been proposed in response to the challenges mentioned earlier. Lactoferrin, a product of PMN, was assessed in ascitic fluid and found to have a high sensitivity of almost 96% at a cut-off value of 51.4 ng/mL. However, its specificity was slightly lower at 74%[16].

Another notable marker was calprotectin, a protein derived from neutrophils. This marker has been proposed as an alternative to the traditional method for diagnosing SBP. It was found to be highly significant at a cut-off value of 1.57 μg/mL, with a sensitivity of 88% and a specificity of 98%[17].

A recent meta-analysis[18] assessed the effectiveness of lactoferrin and calprotectin as diagnostic markers for SBP. The analysis encompassed 13 studies on lactoferrin and 12 studies on calprotectin in ascitic fluid. The researchers concluded that both markers could serve as rapid alternatives to the currently employed techniques. The ratio of calprotectin to total protein has been investigated as a diagnostic and prognostic marker for SBP. A study has shown that elevated levels of this ratio are strongly linked to the diagnosis of SBP, with an impressive area under the curve (AUC) of 0.93 and a sensitivity and specificity of 93% and 79%, respectively[19].

In a recent study, Aehling et al[20] found that the detection rate of bacterial DNA could be enhanced using 16S rRNA PCR compared to traditional cultures. The authors also reported that higher bacterial DNA quantity was associated with poorer prognosis and higher mortality. In the same study, ascitic interleukin-6 (IL-6) levels in patients with SBP were compared to those without SBP. The results showed a strong association with SBP, with an AUC of 0.810, a sensitivity of 78.6%, and a specificity of 83% at a cut-off value of 2.0 × 10⁵ pg/mL.

Additional markers present in ascitic fluid were macrophage inflammatory protein type 1 beta and high-sensitivity C-reactive protein (CRP). The levels of the former were significantly higher in the ascitic fluid of patients with SBP compared to those without SBP. Similarly, the levels of high-sensitivity CRP were elevated in both the serum and ascitic fluid of patients with SBP compared to those without SBP[21].

NON-INVASIVE DIAGNOSIS OF SBP: THE NEXT FRONTIER

Noninvasive markers

As previously mentioned, the traditional method of diagnosing SBP poses a significant challenge owing to its invasive nature. As a result, various markers have been extensively researched in the literature to identify a noninvasive, efficient, and time-saving alternative.

The most extensively studied noninvasive serum marker is procalcitonin. In a large cohort study of 362 individuals with liver cirrhosis and ascites in China, it was found that serum procalcitonin levels were significantly higher in patients with SBP[22]. The researchers concluded that a cutoff value of 0.462 ng/mL had a sensitivity of 84% and a specificity of 95%. A more recent study established that a cutoff value of > 2 ng/mL had an AUC of 0.75, with a sensitivity of 50% and a specificity of 100%[23]. The same study examined the role of serum CRP in diagnosing SBP. Despite its high sensitivity of 100%, the inflammatory marker exhibited a very low specificity of 6.5% at a cut-off level of 3 mg/L. In a separate study, high-sensitivity CRP was found to be significantly higher in both serum and ascitic fluid of patients with SBP[24].

Serum homocysteine was another noninvasive marker for diagnosing SBP, as reported by Abdel-Razik et al[25]. The authors reported that a cut-off value of 17.79 μmol/L had 89.3% specificity and 95.1% sensitivity for distinguishing SBP, with an AUC of 0.93.

Similarly, serum IL-6 has also shown diagnostic value, with an AUC of 0.764. The identified cut-off value was 442.37 pg/mL, providing a sensitivity of 73.8% and specificity of 80.4%. However, it is important to note that while serum IL-6 levels showed an association with SBP, their reliability as a diagnostic marker may not be entirely reliable owing to their involvement in various inflammatory conditions, including malignancies and infections[20].

A study assessing the potential of serum calprotectin as a diagnostic marker for SBP revealed that a cut-off value of ≥ 46.0 µg/mL demonstrated an AUC of 0.976, with a sensitivity of 100% and a specificity of 92%. These findings suggest that serum calprotectin shows promise as a valuable diagnostic tool for identifying SBP[26].

According to the findings of Abudeif et al[27], serum copeptin has been identified as a risk factor for the occurrence of SBP. However, it is worth noting that this biomarker is typically elevated in individuals with advanced liver cirrhosis and associated complications.

Other markers that currently have insufficient evidence to support their use and require further investigation include interferon γ-induced protein 10 (IP-10) and tumor necrosis factor-α. Although both markers were found to be highly sensitive and specific, interferon γ-induced protein 10 was identified to have a higher sensitivity and specificity in both ascitic fluid and serum[28].

Although these noninvasive markers suggest the presence of an ongoing inflammatory process, they cannot be considered highly specific to confirm the diagnosis of SBP, as they are barely indicative of a systemic inflammatory response process.

Role of artificial intelligence and machine learning in SBP diagnosis

The integration of artificial intelligence (AI) and machine learning (ML) in the field of medicine continues to advance. These technologies have proven to be highly beneficial in various areas of hepatology. Numerous studies have explored the use of these technologies to predict SBP (Table 1). A recent development includes a model that uses noninvasive variables such as vital signs, complete blood count, liver and kidney function, blood gases, and potassium levels to effectively rule out the presence of SBP. This model has demonstrated a remarkably high negative predictive value of 95% to 100%, which could significantly reduce the need for paracentesis in settings where it may not be readily available[29].

Table 1 Current models for noninvasive diagnosis of spontaneous bacterial peritonitis.

Ref.	Items
Silvey et al[29]	Vital signs
	Complete blood count
	Liver and kidney function
	Blood gases
	Potassium levels
Würstle et al[30] (Lasso score primary model)	C-reactive protein
	Previous hydropic decompensation
	White blood cell counts
	Organ failure
	Fever
	Acute gastrointestinal bleeding
	Proton pump inhibitors use
	Previous attack
	Charlson comorbidity index
	No propranolol or carvedilol medication
	Model for end-stage liver disease score
Würstle et al[30] (Simple model)	C-reactive protein
	Previous hydropic decompensation
	White blood cell counts
	Organ failure
	Fever
	Acute gastrointestinal bleeding
	Proton pump inhibitors use
	Previous attack
	No propranolol or carvedilol medication
	Child-Pugh class C
Xiang et al[31]	Procalcitonin
	C-reactive protein
	Lymphocyte percentage
	Mean cell hemoglobin concentration
	Total bilirubin
	Albumin
	Prothrombin time
Abdel-Razik et al[25] (Mansoura score)	Age
	Blood neutrophil-lymphocyte ratio
	Mean platelet volume
	C-reactive protein
Tu et al[32]	Model for end-stage liver disease score
	Polymorphonuclear neutrophil
	Blood neutrophil percentage
	Presence of hepatocellular carcinoma
	Renal impairment

Additionally, Würstle et al[30] introduced two new scoring systems (Table 2) under the Lasso Score model. In the primary model, a score exceeding 72 strongly suggested the presence of ascites, with a sensitivity of 94.7% and a specificity of 42.3%, while the simplified model indicated that a score greater than 23 had a sensitivity of nearly 95%.

Table 2 Lasso score model.

Measurement	Primary model	Simple model
CRP > 4.2 mg/dL	+87	+26
Previous hydropic decompensation	+60	+17
White blood cell counts > 11.49 G/L	+52	+16
Organ failure	+45	+14
Fever	+39	+13
Acute gastrointestinal bleeding	+31	+8
PPI medication	+26	+8
Previous SBP	+8	+5
Charlson comorbidity index > 6	+6
No propranolol or carvedilol medication	+1	+2
MELD-Na score > 24.9	+1
Child-Pugh class C		+1

CRP: C-reactive protein; PPI: Proton pump inhibitor; SBP: Spontaneous bacterial peritonitis.

Another scoring system was created by Xiang et al[31], which incorporated seven noninvasive variables. The authors also developed an online calculator for estimating the risk of SBP (https://cqmuxss.shinyapps.io/dynnomappdc_and_sbp/). This model has achieved an AUC of 0.80, with a sensitivity of 75.5% and specificity of 72.9% for a cutoff value of −1.514.

The Mansoura scoring system is considered one of the simplest available scoring systems (Table 3), which incorporates age, blood neutrophil-lymphocyte ratio, mean platelet volume, and CRP. While the system demonstrates high specificity (97.8) and a positive predictive value of 88.1% for scores ≥ 4, its sensitivity is relatively low at 55%[25].

Table 3 Mansoura scoring system.

Parameter	Cut-off	Score
Age (years)	≥ 55	1
NLR	≥ 2.5	1
MPV (femtoliters)	≥ 8.5	1
CRP (mg)	≥ 40	2

NLR: Neutrophil-lymphocyte ratio; MPV: Mean platelet volume; CRP: C-reactive protein.

The diagnosis of asymptomatic SBP remains a challenge in clinical practice. To address this, Tu et al[32] introduced a predictive model for diagnosing SBP in asymptomatic patients. The model incorporated the Model for End-stage Liver Disease score, PMN, blood neutrophil percentage, presence of hepatocellular carcinoma, and renal impairment. It demonstrated a sensitivity of 73.5%, a specificity of 86.7%, and an AUC of 0.872.

Despite the aforementioned use of artificial intelligence and noninvasive methods in the diagnosis of SBP, several challenges, such as ethical considerations, data security, and privacy, must be addressed before clinicians can implement these technical improvements in a clinical setting. Additionally, the limitations of such technologies in low-resource settings could delay their use in real-world clinical practice.

Future research directions and collaboration

As long as technological and methodological advancements continue to develop, research in the diagnosis of SBP has the potential to improve and become less invasive. Emphasizing early detection and accuracy, current and future studies focus on identifying new biomarkers and better imaging techniques to develop less invasive yet accurate diagnostic methods. Furthermore, the use of artificial intelligence and machine learning in assessing clinical data is expected to lead to more effective identification of at-risk patients and more tailored therapeutic methods. Standardizing diagnostic criteria and ensuring the consistent application of best practices will require cooperative efforts throughout the medical community to reduce death rates associated with this SBP. Moving forward, the combination of innovative research and clinical practice will be crucial to enhance patient outcomes and rethink treatment approaches for SBP.

CONCLUSION

Despite the consistent demonstration of high sensitivity of diagnostic paracentesis and PMN counting, efforts have been directed toward improving these invasive methods for diagnosing SBP. We believe the focus should now shift toward developing a noninvasive rapid test for diagnosing SBP. While diagnostic models using machine learning may be useful in settings where paracentesis is unavailable, they often lack specificity or sensitivity and may not effectively assess response to treatment, especially in asymptomatic patients. Hence, further studies are required to introduce more advanced noninvasive diagnostic tests for SBP. Such tests would be particularly beneficial if they could be utilized for post-treatment follow-up.