Role of non-invasive serological and CT imaging parameters in assessing portal hypertensive gastropathy severity

Abstract

Introduction

To compare serological and CT imaging parameters between the mild and severe portal hypertensive gastropathy (PHG) group, and find out the non-invasive predictive markers for PHG severity based on these differences.

Methods

Three hundred one patients were enrolled in our retrospective study. The PHG severity was assessed by esophagogastroduodenoscopy (EGD), and serological markers were measured within 48 h before EGD, and CT scans were performed within one week after EGD. Differences in age, gender, etiology, serological and CT imaging parameters, and arterial-phase gastric wall enhancement in CT scans were analyzed between the mild and severe PHG groups.

Results

No significant differences were observed in gender, age, etiology, serological markers, or MELD score between the two groups (all P > 0.05). However, spleen length, spleen volume, PLT/spleen length ratio, and PLT/spleen volume ratio differed significantly (P = 0.003, 0.011, 0.019, and 0.014, respectively). The optimal cutoff values for predicting PHG severity were 131.4 (AUC = 0.619), 613.29 (AUC = 0.596), 0.42 (AUC = 0.588), and 0.07 (AUC = 0.592), respectively. Patients with severe PHG demonstrated significantly greater arterial-phase gastric wall enhancement compared to mild PHG cases.

Conclusion

Spleen length, spleen volume, and the platelet-to-spleen ratios presented predictive value for PHG severity, albeit with limited discriminative ability. Additionally, moderate-to-severe arterial phase gastric wall enhancement in CT scans may serve as a potential indicator of severe PHG.

Introduction

Portal hypertensive gastropathy (PHG), a complication of decompensated cirrhosis with a reported incidence ranging from 20 to 98% [1], predominantly affects the gastric body and fundus [2]. Histopathological examination reveals characteristic findings: punctate or diffuse hemorrhages on light microscopy, and tortuous, dilated submucosal vessels on electron microscopy, with or without inflammatory changes [2]. As a significant etiology of upper gastrointestinal bleeding in cirrhotic patients, PHG accounts for approximately 10%, only secondary to esophageal and gastric varices [3]. PHG is often overlooked clinically but can be life-threatening if a massive hemorrhage occurs [4]. Advances in endoscopy have improved its detection and recognition.

Multiple studies have established an association between PHG and portal hypertension [1, 5, 6]. In cirrhotic patients, portal hypertension not only causes esophageal and gastric varices but also alters gastric mucosal hemodynamics. This results in hyperdynamic congestion, triggering cytokine and growth factor activation. These pathological changes compromise the gastric mucosal barrier through pro-inflammatory mediator release and growth factor suppression, ultimately leading to mucosal damage, erosion, ulceration, and hemorrhage [7]. However, the determinants of PHG severity remain controversial [6, 8–10], and the correlation between CT imaging parameters and PHG severity has been scarcely investigated.

Currently, esophagogastroduodenoscopy (EGD) remains the gold standard for PHG diagnosis [4]. However, its invasive nature and variable patient acceptance-particularly in those with cardiac or coagulation dysfunction-highlight the need for reliable non-invasive alternatives. Our study retrospectively analyzes correlations between PHG severity and: (1) serological markers, (2) CT imaging parameters, and (3) arterial-phase gastric wall enhancement patterns, aiming to identify non-invasive predictive indicators.

Materials and methods

Subjects

This study included 324 cirrhotic patients with PHG confirmed by EGD from our gastroenterology department (June 2015-December 2023). Our study was approved by the institutional review board with a waiver of informed consent. They completed EGD, serological, and CT data. Cirrhosis was confirmed clinically, radiologically or histologically, with documented etiology. Exclusion criteria: (1) severe cardiopulmonary dysfunction precluding EGD cooperation; (2) exhibited splenomegaly from non-cirrhotic causes; (3) combined with hepatocellular carcinoma or chronic renal disease; (4) underwent prior esophageal variceal ligation.

From the initial 324 patients, 23 were excluded due to incomplete (n = 11) or untimely (> 48 h before EGD, n = 12) serological tests. The remaining 301 eligible patients (age ranged 13–78 years; males 13–78, females 17–77; mean age 49.79 ± 11.37 years; males 47.92 ± 10.73, females 57.3 ± 10.8) underwent further evaluation, with 64 excluded from CT analysis due to incomplete spleen coverage (n = 55), non-enhanced scans (n = 3), or respiratory artifacts (n = 6), yielding 237 cases with complete serological and quality CT data (Fig. 1).

Fig. 1.

The flow diagram of the current study

EGD examination

All EGD procedures were performed by a senior endoscopist (14 years of experience) using an Olympus CV-290 system. Examinations were conducted on patients with stable conditions after a standardized 12-h fasting period. Classification and scoring of PHG were performed according to the Baveno II scoring system [11]. A mild mucosal mosaic pattern was assigned 1 point, and a severe pattern received 2 points; isolated red markings were given 1 point, while confluent markings were assigned 2 points. The presence of gastric antral vascular ectasia (GAVE) was also noted, adding 2 points to the score when flat or slightly raised red stripes were observed radiating from the pylorus to the antrum and body of the stomach. Based on the total score, PHG was classified as mild (score ≤ 3) or severe (score ≥ 4).

Clinical serological markers

Laboratory parameters, including liver function tests (ALT, AST, GGT, ALP, TB, IB, ALB), coagulation profile (PT, PTA, APTT, INR), and platelet count (PLT), were collected within 48 h before EGD. The MELD score was calculated using standard laboratory values with the following formula: .

CT imaging parameters analysis

Two abdominal radiologists (11 and 8 years of experience) independently analyzed the contrast-enhanced CT images. They were blind to the clinical and serological data. Measurements included: spleen dimensions (length/width/thickness), portal/splenic/superior mesenteric vein diameters. The following definitions are provided for the measurement of spleen dimensions:

Spleen length: The maximum distance between the superior and inferior margins of the spleen on a coronal plane passing through the splenic hilum. Spleen width: The distance between the medial and lateral edges of the spleen on a plane perpendicular to its maximum thickness. Spleen thickness: The maximum anteroposterior diameter of the spleen measured in an axial cross-section. Derived parameters comprised spleen volume (calculated as [spleen volume(mL) = 30 + 0.58 × spleen length(mm) × spleen width(mm) × spleen thickness(mm)/1000]) [12], PLT/spleen length ratio, and PLT/spleen volume ratio.

Meahwhile, arterial-phase gastric wall enhancement (graded as no-mild or moderate-severe) were recorded. Enhancement of the gastric wall in the arterial phase is categorized into three grades: mild, defined as an increase in MSCT attenuation value of ≤ 20 HU; moderate, defined as an increase of > 20 HU but ≤ 40 HU; and severe, defined as an increase of > 40 HU.

Statistics

Statistical analyses were performed using SPSS 21.0 and GraphPad Prism 10.4.1. Normally distributed continuous variables were expressed as mean ± standard deviation (SD), while nonparametric data were reported as median (interquartile range). Intergroup comparisons employed Student’s t-test or Mann–Whitney U test for continuous variables, and χ² or Fisher’s exact test for categorical variables (presented as n%). Diagnostic performance was assessed through ROC analysis (reporting AUC, sensitivity, specificity, and optimal cutoff values). All tests were two-tailed, with P < 0.05 considered statistically significant.

Results

Association between PHG severity and clinical/serological parameters

The cohort comprised 301 patients with PHG: 153 with mild PHG (117 males, 36 females; mean age 49.56 ± 11.19 years) and 148 with severe PHG (120 males, 28 females; mean age 49.87 ± 11.17 years). No significant differences were observed in age (P = 0.806) or sex distribution (P = 0.328) between groups (Table 1).

Table 1.

Clinical and serological characteristics of PHG severity groups

Parameters	PHG			P
	Mild PHG (n = 153)	Severe PHG (n = 148)	All (n = 301)
Sex				0.328
Male	117 (38.87%)	120 (39.87%)	237 (78.74%)
Female	36 (11.96%)	28 (9.30%)	64 (21.26%)
Age, year	49.56 ± 11.19	49.87 ± 11.17	49.71 ± 11.16	0.806
Etiology				0.243
HBV	97 (32.23%)	97 (32.23%)	194 (64.45%)
Alcoholic	16 (5.32%)	18 (5.98%)	34 (11.30%)
HCV	11 (3.65%)	11 (3.65%)	22 (7.31%)
Autoimmunity	3 (1.00%)	1 (0.33%)	4 (1.33%)
Autoimmunity + Primary biliary	2 (0.66%)	0 (0%)	2 (0.66%)
Cholestatic	6 (1.99%)	1 (0.33%)	7 (2.33%)
Cryptogenic	10 (3.32%)	8 (2.66%)	18 (5.98%)
Drug-induced	1 (0.33%)	0 (0%)	1 (0.33%)
Essential thrombocythemia	0 (0%)	1 (0.33%)	1 (0.33%)
HBV + Alcoholic	1 (0.33%)	4 (1.33%)	5 (1.66%)
HBV + HCV	1 (0.33%)	0 (0%)	1 (0.33%)
HCV + Alcoholic	3 (1.00%)	1 (0.33%)	4 (1.33%)
Hepatolenticular degeneration	0 (0%)	2 (0.66%)	2 (0.66%)
Liver fluke	1 (0.33%)	0 (0%)	1 (0.33%)
Myelofibrosis	0 (0%)	1 (0.33%)	1 (0.33%)
Portal hypertension	1 (0.33%)	0 (0%)	1 (0.33%)
Portal venous phlebangioma	0 (0%)	1 (0.33%)	1 (0.33%)
Sclerosing cholangitis	0 (0%)	2 (0.66%)	2 (0.66%)
Serological markers
ALT (U/L)	26 (20, 47)	30 (22, 40.75)	28 (21, 42)	0.508
AST (U/L)	37 (27, 57.5)	36 (28, 47)	37 (28, 53.5)	0.387
GGT (U/L)	41 (23, 117)	42.5 (26.25, 89.5)	42 (26, 92)	0.719
ALP (IU/L)	74 (56.5, 98.5)	80.5 (58.25, 108.75)	76 (57, 105)	0.237
TB (umol/L)	18.4 (11.8, 29.45)	17.01 (11.83, 26.4)	17.8 (11.85, 27.15)	0.595
IB (umol/L)	10.7 (6.8, 15.3)	9.7 (6.63, 14.88)	10.1 (6.7, 15.05)	0.547
ALB (g/L)	34.58 ± 5.41	34.65 ± 5.13	34.62 ± 5.26	0.907
PTA (%)	67.74 ± 15.51	67 (59.25, 76)	68 (58, 77)	0.977
PT (sec)	15.9 (14.8, 17.7)	16 (15.03, 17.28)	16 (14.95, 17.5)	0.782
APTT (sec)	39.4 (35.9, 43.5)	39.95 (37.45, 43.65)	39.6 (36.55, 43.55)	0.332
PT-INR	1.27 (1.17, 1.46)	1.3 (1.19, 1.42)	1.28 (1.18, 1.44)	0.892
PLT (10E9/L)	79 (56, 106)	74 (53, 110.25)	78 (54, 107)	0.656
MELD	10.24 (8.67, 12.7)	10.24 (8.86, 12.19)	10.24 (8.72, 12.32)	0.844

Etiological analysis of the 301 cirrhotic patients revealed no significant difference in underlying causes between mild and severe PHG groups (P = 0.243, Table 1). Similarly, comparison of serological markers (presented as mean ± SD or median [interquartile range]) and MELD scores showed no statistically significant intergroup differences (all P > 0.05, Table 1).

Correlation between PHG severity and CT imaging parameters

Table 2 presents the median (interquartile range) or mean ± SD values of splenic dimensions (length/width/thickness), portal/splenic/superior mesenteric veins, and derived indices (spleen volume, PLT/spleen length ratio, PLT/spleen volume ratio) for the 237 patients. Among 237 patients with analyzable CT data, spleen length (P = 0.003) and volume (P = 0.011) were significantly greater in severe versus mild PHG, otherwise PLT/spleen length ratio (P = 0.019) and PLT/spleen volume ratio (P = 0.014) were significantly lower (Table 2, Fig. 2). No significant differences were observed in other splenic or venous measurements (all P > 0.05).

Table 2.

CT imaging parameters of PHG severity groups

Parameters	PHG		P
	Mild PHG (n = 121)	Severe PHG (n = 116)
Spleen length	131.28 ± 24.02	140.28 ± 22.79	0.003
Spleen width	58.40 (50.00, 68.90)	62.36 ± 11.78	0.077
Spleen thickness	144.63 ± 29.99	151.91 ± 28.03	0.055
Portal vein diameter	16.19 ± 2.84	16.6 (14.83, 18.48)	0.090
Splenic vein diameter	11.62 ± 3.17	11.9 (9.8, 13.78)	0.430
Superior mesenteric vein diameter	12.92 ± 2.13	13.22 ± 2.07	0.267
Spleen volume	650.77 (471.73, 952.89)	760.19 (591.59, 1047.08)	0.011
PLT/spleen volume	0.123 (0.076, 0.202)	0.097 (0.058, 0.164)	0.014
PLT/spleen length	0.624 (0.478, 0.893)	0.541 (0.372, 0.768)	0.019

Fig. 2.

Comparison of splenic and PLT-based ratios between PHG severity groups. A Spleen length (P = 0.003). B Spleen volume (P = 0.011). C PLT/spleen length ratio (P = 0.019). D PLT/spleen volume ratio (P = 0.014)

Arterial-phase gastric wall enhancement patterns were detailed in Table 3. Patients with severe PHG demonstrated significantly greater enhancement compared to mild PHG (Figs. 3 and 4).

Table 3.

Arterial-phase gastric wall enhancement in PHG severity groups

	Arterial-phase gastric wall enhancement		Total	P
	No-mild	Moderate-severe
The mild PHG group	102 (84.3%)	19 (15.7%)	121	0.008
The severe PHG group	81 (69.8%)	35 (30.2%)	116

Fig. 3.

Three patients showed mild PHG: a 57-year-old man (5-year cirrhosis), a 53-year-old woman (7-year), and a 47-year-old man (5-year). A, D, G EGD reveals an isolated mosaic pattern without red markings in the gastric body, consistent with mild portal hypertensive gastropathy (PHG). On arterial-phase CT (C, F, I), mild gastric wall enhancement is observed, with attenuation increases of 3 HU, 6 HU, and 2 HU compared to pre-contrast images (B, E, H)

Fig. 4.

Three patients showed severe PHG: a 52-year-old man (6-year cirrhosis), a 50-year-old man (5-year), and a 44-year-old man (8-year). A, D, G EGD revealed severe PHG in the gastric body, exhibiting diffuse cherry-red spots. On arterial-phase CT (C, F, I), severe gastric wall enhancement is observed, with attenuation increases of 42 HU, 66 HU, and 43 HU compared to pre-contrast images (B, E, H)

ROC analysis evaluated the diagnostic performance of the indicators for PHG severity

CT imaging parameters displayed limited discriminative capacity for PHG severity stratification (AUC range: 0.588–0.619). Spleen length demonstrated the highest accuracy (AUC = 0.619, sensitivity = 67.24%, specificity = 57.85%, cutoff > 131.4 mm, P = 0.001), followed by spleen volume (AUC = 0.596, sensitivity = 73.28%, specificity = 46.28%, cutoff > 613.29 mL, P = 0.009). PLT-based ratios exhibited higher specificity (PLT/spleen length ratio: 85.12%; PLT/spleen volume ratio: 78.51%) but lower sensitivity (33.62–35.34%) at their respective cutoffs (both P < 0.05) (Fig. 5).

Fig. 5.

ROC curves of CT imaging parameters. A Spleen length. B Spleen volume. C PLT/spleen length ratio. D PLT/spleen volume ratio for PHG severity prediction

Discussion

PHG may result from cirrhotic or non-cirrhotic portal hypertension, leading to hyperdynamic circulation and gastric congestion [13]. While its exact pathogenesis remains unclear, proposed mechanisms include elevated oxygen free radicals, lipid peroxidation, TNF-α, and glucagon, increasing mucosal injury susceptibility [14–17]. Both research and our center's experience indicate that severe PHG carries a high bleeding risk (> 90%) due to fragile, dilated gastric vessels. This can lead to severe hemorrhage or shock, necessitating aggressive treatments like argon plasma coagulation, TIPS, or liver transplantation, while non-selective beta-blockers (NSBBs) are also required to prevent recurrent bleeding [16]. These patients require close follow-up to adjust therapy and prevent rebleeding. In contrast, mild PHG has a low bleeding risk (< 10%). Primary prophylaxis is not recommended; management typically involves NSBBs to reduce portal pressure, along with regular endoscopic monitoring [16]. The prognosis is generally good. Severe PHG is associated with significantly shorter survival compared to mild or no PHG [1, 18]. In our study, no differences in gender, age, or etiology were observed between mild and severe PHG groups. However, spleen length, spleen volume, PLT/spleen length ratio, and PLT/spleen volume ratio showed statistically significant differences. While these four indicators can predict PHG severity, their predictive value remains limited. Though portal hypertension is essential for PHG development [17], it alone cannot explain PHG occurrence, as many patients with portal hypertension remain unaffected [19, 20]. Portal vein pressure increases correlate with dilation of the portal, splenic, and superior mesenteric veins. Studies showed different results on PHG-vein diameter correlations: Kim MY, et al. found PHG severity correlated with portal/splenic vein diameters [18], El-Kalla F, et al. found larger vein diameters in PHG patients than in non-PHG patients [21], Zardi EM, et al. noted no PHG degree association, but significant portal vein diameter differences between cirrhotic with/without PHG [22]. This study found no significant difference in vessel diameters between mild and severe PHG groups, suggesting PHG differs from simple gastric varices and may involve gastric mucosal factors [7, 8].

Portal hypertension and severe liver disease are key predictors of PHG [16, 23]. Though Zardi EM, et al. found no significant splenic diameter differences between mild and severe PHG [22], Nishino K, et al. and El-Kalla F, et al. identified that the size of the spleen was related to the occurrence and development of PHG [3, 21]; meanwhile, Nishino K, et al. and Min YW, et al. proposed that spleen size was an independent risk factor for predicting PHG [3, 24]. A prospective study of 331 patients with liver cirrhosis in South Korea stated that PHG severity was correlated with splenic diameter [18]. Our findings corroborate this observation, showing significantly greater spleen longest diameters in severe versus mild PHG cases.

PHG severity was negatively correlated with platelet count [16, 25]. In cirrhotic portal hypertension, increased splanchnic inflow results in splenomegaly and subsequent thrombocytopenia [26]. In patients with chronic liver disease, thrombocytopenia can arise from multiple mechanisms, including portal hypertension, reduced platelet survival, diminished thrombopoietin production, and myelotoxic effects of alcohol or viral hepatitis. In cirrhotic patients, splenomegaly frequently results from hemodynamic alterations secondary to portal hypertension. The PLT/spleen length and PLT/spleen volume ratios thus integrate thrombocytopenia with splenic enlargement, providing a quantitative indicator of hypersplenism-related thrombocytopenia due to portal hypertension [27]. Several studies have validated these ratios as noninvasive tools for predicting the severity of esophageal varices in patients with liver cirrhosis [28–32]. Mandhwani R, et al. proposed that the platelet count to spleen diameter ratio (PSR) was a relatively better predictor of PHG but showed a lower specificity and negative predictive value (NPV) [4]. This is consistent with our results. Elevated portal pressure impedes gastric venous drainage, leading to vascular dilation, congestion, and arteriovenous shunting in the gastric mucosa and submucosa. This reduces mucosal perfusion, resulting in ischemic and hypoxic injury [8]. PHG alters the gastric microcirculation, but the mucosal blood flow remains debated [16, 33]. Most studies reported reduced mucosal perfusion, but increased blood flow was observed in submucosal/muscular layers and total gastric circulation [33, 34]. This study found more frequent arterial-phase gastric wall enhancement in severe PHG patients. We hypothesize that reduced mucosal blood flow in early PHG yields less arterial enhancement. As PHG progresses, mucosal hypoxia and elevated vascular endothelial growth factor (VEGF)/proliferating cell nuclear antigen (PCNA) levels may promote angiogenesis, ultimately increasing blood flow [35]. Meanwhile, transient gastric perfusion defects serve as a specific sign for PHG in cirrhotic patients [36], supporting the hypothesis of reduced gastric mucosal blood flow in PHG [37].

So far, the association between PHG severity and demographic/serological factors remains controversial. Simbrunner B, et al. proposed that higher male predominance was reported in severe PHG [38]. In addition, gender, prothrombin time, and albumin, but not age or etiology, were found to be associated with PHG development [8]. Wang W, et al. considered age, hemoglobin, and platelet count as independent PHG risk factors [13]. Whereas Zardi EM, et al. found no significant age or gender differences between mild/severe PHG groups [39]. Multiple studies consistently demonstrated that cirrhosis etiology did not influence PHG development [6, 18, 40–43]. Iwao T, et al. further confirmed PHG’s independence from age, sex, or etiology [44], which aligns with our findings.

Though FIB-4, APRI, ALBI, AAR, GPR, and MELD scores were widely used in the hepatic fibrosis staging [45], they were rarely applied in PHG prediction. Our previous findings demonstrated that FIB-4, APRI, ALBI, AAR, and GPR had minimal to no predictive value for PHG [46]. Existing studies reported conflicting results regarding MELD score and PHG severity. While some studies found no significant differences across PHG severity groups [4, 22, 38], which is consistent with our findings. Several data identified PLT and ALB as independent predictors of PHG [24, 47]; meanwhile, others reported significant differences in these markers between mild/severe PHG groups [48]. However, our study found no significant serological differences, potentially due to regular patient monitoring and long-term medication use in our cohort. The relationship between PHG severity and these serological markers requires further prospective investigation.

Our study has several limitations. First, its single-center, retrospective design may introduce selection bias. Second, the sample size (particularly for CT imaging parameters) is insufficient. Finally, complete Child–Pugh score data were unavailable due to the retrospective nature of the data collection. Future prospective, multicenter studies are needed to better characterize the relationships between PHG severity and both serological and CT imaging parameters, which would enhance clinical decision-making.

In conclusion, certain CT imaging parameters, including spleen length, spleen volume, and PLT-based ratios, show limited predictive capability. Therefore, EGD remains necessary in clinical practice to accurately distinguish between severe and mild PHG. Additionally, moderate-to-severe arterial phase enhancement of the gastric wall on CT may serve as a potential indicator of severe PHG, though this finding requires further validation.

Abbreviations

CT

Computed Tomography

PHG

Portal hypertensive gastropathy

EGD

Esophagogastroduodenoscopy

MELD

Model for End-Stage Liver Disease

PLT

Platelet

ALT

Alanine aminotransferase

AST

Aspartate aminotransferase

GGT

Gamma-Glutamyl Transferase

ALP

Alkaline Phosphatase

TB

Total bilirubin‌

IB

Indirect Bilirubin‌

ALB

Albumin

PT

Prothrombin Time

PTA

Prothrombin time activity

APTT

Activated partial thromboplastin time

INR

International normalized ratio

SD

Standard deviation

TNF-α

Tumor Necrosis Factor Alpha

ROC curve

Receiver Operating Characteristic curve

AUC

Area Under Curve

PSR

Platelet count to spleen diameter ratio

NPV

Negative predictive value

VEGF

Vascular endothelial growth factor

PCNA

Proliferating cell nuclear antigen

FIB-4

Fibrosis-4 Index

APRI

Aspartate aminotransferase to platelet ratio index

ALBI

Albumin-bilirubin index

AAR

Aspartate aminotransferase/Alanine aminotransferase ratio

GPR

Gamma-Glutamyl Transferase/ Platelet ratio

Authors’ contributions

Siwei Tan and Jie Qin planned and designed the study. Zhenzhen Liu, Bilun Ke, and Shuyan Tan collected and analyzed the data. Zhicheng Zhang and Dandan Shan drafted the manuscript. All authors have approved the submitted manuscript.

Funding

This work was supported by grants from the National Natural Science Foundation of China (82170569, 82470589); the Science and Technology Planning Projects of Guangzhou City (2025A03J3193); the major talent project training program of the Third Affiliated Hospital of Sun Yat-sen University (P02089).

Data availability

Data and materials generated or analysed during this study are included in this published article, however the raw data are available from the corresponding author on reasonable request.

Declarations

Ethics approval and consent to participate

This study protocol was approved by the Institute Research Ethics Committee of the Third Affiliated Hospital of Sun Yat-sen University (No. II2025-341–01). In our study, participants were exempt from the requirement of obtaining informed consent.

The study is adherent to the declaration of Helsinki.

Consent for publication

All patients included in this research gave written informed consent to publish the data contained within this study upon reasonable request to the corresponding author.

Competing Interests

The authors declare no competing interests.