This retrospective investigation was performed at the Gastroenterology Department of West China Hospital between January 2013 and January 2024. Initially, patients with positive plasma EBV DNA loads and intestinal diseases were screened through electronic medical record system, followed by verification through the pathology system to confirm whether they had also undergone intestinal EBER-ISH test. Patients with intestinal diseases who tested positive for peripheral blood EBV DNA and completed the intestinal EBER-ISH test were included in the study, while those who did not complete both tests, had incomplete clinical data, or had unclear diagnoses were excluded. Immunomodulator usage was defined as treatment with steroids, immunosuppressants, or biological agents within three months prior to the intestinal EBER-ISH test, and intestinal EBV infection was defined as positive EBER-ISH test in the intestine. The diagnosis of IBD followed recognized criteria, including typical clinical presentations, endoscopic findings, radiological assessments, and histological results (Gomollón et al., 2017; Magro et al., 2017). The diagnostic criteria for PILPDs require the presence of gastrointestinal symptoms and confirmation of abnormal proliferation of lymphocytes within the intestinal tissues through pathology, while excluding primary lymphoproliferative diseases at other sites (Wang Z. et al., 2018). On this basis, a positive result for intestinal EBER-ISH test was defined as PIEBV+LPDs, while a negative result was defined as primary intestinal non-EBV-associated lymphoproliferative diseases (PINEBV+LPDs). Data from the electronic medical record system were also collected to analyze clinical data, such as sex, age, plasma EBV DNA load, and other factors. Furthermore, a follow-up on the prognosis of all patients was conducted, where events such as intestinal resection or death within six months after undergoing intestinal EBER-ISH test were defined as fatal events, while all other outcomes were considered benign events. Ethics approval for the present investigation was obtained from the ethics board of West China Hospital. (Number: 2023-22).

As directed by the manufacturer, quantitative measurement of EBV DNA in the plasma was performed using an EBV detection kit with polymerase chain reaction (PCR) fluorescence (Shen Xiang Gene Co.). The EBER gene was amplified using a Bio-Rad CFX96 PCR instrument. A Ct value ≤39 was interpreted as positive, and a standard curve was used to calculate the quantity of EBV DNA copies. EBER-ISH was used to investigate intestinal EBV infection, and the EBV-encoded small RNAs (EBER) peptide nucleic acid probe was sourced from ZSGB-BIO (China). Specifically, tissues from intestinal biopsies of patients were deparaffinized, rehydrated, and permeabilized with proteinase K, followed by overnight hybridization at 37°C with digoxigenin-labeled EBER probes. After incubation with horseradish peroxidase-conjugated anti-digoxigenin antibody, EBER-positive cells were identified by diaminobenzidine staining and exhibited brown-stained nuclei. These cells were quantitatively scored by two experienced pathologists in high-power field (HPF).

Statistical evaluation was performed using SPSS and GraphPad Prism (version 26.0 and 9.5, respectively). This investigation of the demographic and basic features of the patients utilized both frequency distributions and descriptive statistical methods. To compare patient characteristics among different cohorts, either the chi-squared hypothesis evaluation or the exact significance test using Fisher’s exact test was employed, contingent on the distributional features inherent in these data. To evaluate the role of plasma EBV DNA load in the diagnostic and prognostic assessment of intestinal EBV infection, a stepwise analysis was conducted using both the train and validation cohorts. In the first stage, receiver operating characteristic (ROC) curves were utilized to analyze plasma EBV DNA concentrations for diagnosing intestinal EBV infection and predicting its prognosis. These optimal cutoff values were determined based on the maximum Youden’s index, which maximizes the sum of sensitivity and specificity (Youden’s index = sensitivity + specificity-1). This approach ensures the identification of the threshold that achieves the best trade-off between true-positive and true-negative rates for each cohort. In the next stage, the performance of these cutoff values was assessed in the validation cohort by calculating sensitivity, specificity, positive predictive value (PPV), negative predictive value (NPV), positive likelihood ratio (LR⁺), and negative likelihood ratio (LR^-). These metrics were used to confirm the robustness of the cutoff values derived from the train cohort. Statistical significance was represented by a p-value of < 0.05.

Our investigation began with the identification of 213 patients diagnosed with intestinal diseases and tested positive for plasma EBV DNA. After excluding patients with unclear diagnoses and incomplete clinical data, the remaining patients were randomly assigned in a 2:1 ratio, resulting in a cohort of 108 patients in the train group and 56 patients in the validation group ( Supplementary Figure 1 ; Supplementary Table 1 ). The train cohort consisted of 60 patients (55.56%) who tested negative for intestinal EBER-ISH and 48 patients (44.44%) who tested positive for intestinal EBER-ISH. The median age of the EBER-negative group was 42 years, which was comparable to that of the EBER-positive group (median age, 44 years). Moreover, neither sex distribution nor disease course demonstrated notable differences between the two groups. Additionally, the EBER-positive group did not exhibit a significantly higher frequency of immunomodulator use within three months preceding the intestinal EBER-ISH test than the EBER-negative group. However, a disparity in disease type was observed between the two groups. The EBER-positive group predominantly consisted of patients with PILPDs (50%), followed by IBD (47.9%) and other diseases (2.1%), whereas IBD was the most prevalent disease in the EBER-negative group (76.7%). The symptoms in both groups also differed significantly in clinical terms, with the EBER-positive group showing a higher incidence of fever and hematochezia than the EBER-negative group ( Table 1 ).

In this study, we conducted comparative analyses of the distribution of plasma EBV DNA load across various intestinal diseases. The findings revealed that the majority of patients (88.3%) in the EBER-negative group had EBV DNA loads <10¹ copies/ml, with only 11.7% having loads of 10²-10³ copies/ml, whereas the EBER-positive group exhibited a wider distribution of higher plasma EBV DNA loads, with 33.3% having loads >10³ copies/ml and 25% having loads between 10² and 10³ copies/ml (P < 0.001; Figure 1A ). Additionally, when focusing on patients with IBD, we observed that EBER-positive IBD patients exhibited a notably elevated proportion of plasma EBV DNA loads (10²-10³ copies/ml) compared to their EBER-negative counterparts (30.4% vs. 6.5%, P = 0.013; Figure 1B ). Among patients with PILPDs, 66.6% of PIEBV+LPD patients had EBV DNA loads >10³ copies/ml, whereas no patients in the PINEBV+LPD group exhibited similarly high viral loads (P = 0.001; Figure 1C ). Despite the overall high occurrence of elevated plasma EBV DNA loads in the EBER-positive IBD and PIEBV+LPD groups, a notable difference was observed in the distribution of viral loads between these two groups. In the PIEBV+LPD group, 66.6% of the patients had EBV DNA loads >10³ copies/ml, whereas none of the EBER-positive IBD patients were present within this viral load range (P < 0.001; Figure 1D ).

Further analysis was conducted to compare plasma EBV DNA levels across EBER-negative and EBER-positive populations. Among the 48 patients in the intestinal EBER-positive group, the plasma EBV DNA level had a median value of 2.02 × 10² copies/mL (Interquartile Range [IQR]: 5.49×10¹-6.34×10³ copies/mL), which represented a significantly elevated level compared to the median of 4.2×10¹ copies/mL (IQR: 1.07×10¹-6.08×10¹ copies/mL) observed in 60 patients in the intestinal EBER-negative group ( Figure 1E ). These median plasma EBV DNA loads were notably elevated in comparison to the respective EBER-negative control groups, including the IBD and PILPD groups, as detailed in Figures 1F, G . Notably, within the intestinal EBER-positive cohort, patients diagnosed with PIEBV+LPDs demonstrated the highest median plasma EBV DNA level at 6.18×10³ copies/mL (IQR 1.8×10²-5.91×10⁴ copies/mL), significantly surpassing the levels in EBER-positive IBD patients (5.96×10¹ copies/mL [IQR 2.23×10¹-2.45×10² copies/mL]), with a P-value of < 0.0001 ( Figure 1H ).

According to the consensus established on tissue identification and diagnostic pathology of intestinal EBV infection from the Chinese Medical Association (Ye et al., 2019), we divided these patients into three categories based on the number of cells positive for EBER-ISH per HPF: <10 cells per HPF, 10-50 cells per HPF, and >50 cells per HPF. We analyzed the interrelationship between plasma EBV DNA load and the frequency of intestinal EBER-positive cells per HPF in the intestinal EBER-positive cohort and found that the largest segment comprised patients with <10 cells positive for EBER-ISH per HPF (50%), followed by those with 10-50 cells positive for EBER-ISH per HPF (31.3%), and then patients with >50 cells positive for EBER-ISH per HPF (18.7%; Figure 2A ). Notably, there was a notable disparity between PIEBV+LPDs and EBER-positive IBD, with 79.2% of PIEBV+LPD patients exhibiting more than 10 cells positive for EBER-ISH per HPF, compared to only 17.4% in the EBER-positive IBD group (P < 0.001; Figure 2A ).

For a profound evaluation of EBV DNA load in plasma within intestinal EBV infection, we assessed the median plasma EBV DNA concentrations across different count ranges of EBER-positive cells per HPF, and we discovered that the median plasma DNA concentrations of EBV in the group with >50 EBER-positive cells per HPF (6.41×10⁴copies/mL) and in the group with 10-50 EBER-positive cells per HPF (4.62×10² copies/mL) were both markedly elevated compared to the level within the group with <10 EBER-positive cells per HPF (5.53×10¹ copies/mL), as shown in Figure 2B . Similarly, within the EBER-positive IBD subgroup, patients with 10-50 EBER-positive cells per HPF demonstrated significantly elevated plasma EBV DNA levels compared to those with <10 EBER-positive cells per HPF (3.54×10² copies/mL vs. 5.62×10¹ copies/mL; Figure 2C ). Furthermore, in PIEBV+LPD group, patients with >50 EBER-positive cells per HPF exhibited markedly higher plasma EBV DNA levels (median 6.41×10⁴ copies/mL), which were much superior in comparison with those with 10-50 EBER-positive cells per HPF (median 3.52×10³ copies/mL, P = 0.0076) and those with <10 EBER-positive cells per HPF (median 5.09×10¹ copies/mL, P = 0.0127), as delineated in Figure 2D .

Considering that the density of intestinal cells positive for EBER-ISH per HPF exhibited a positive correlation with plasma EBV DNA levels, receiver operating characteristic (ROC) curve examination was initiated to ascertain the potency of plasma EBV DNA load for screening intestinal EBV infection in the test cohort. The analysis indicated that the area under the ROC curve (AUC) for distinguishing intestinal EBER positivity from EBER negativity was 0.82 at a concentration of 9.21×10¹ copies/mL of EBV DNA in plasma ( Figure 3A ), with a sensitivity of 64.58% and a specificity of 88.33%, indicating its potential utility in initial screening for EBV status. Furthermore, subgroup analyses stratified by intestinal EBER-ISH status identified 5.28×10¹ copies/mL as the cutoff value for differentiating EBV infection from non-EBV infection in IBD, yielding an AUC of 0.73 ( Figure 3B ), with a sensitivity of 69.57% and a specificity of 71.74%. This result suggests that plasma EBV DNA load has moderate discriminatory power in distinguishing EBV-positive IBD from EBV-negative IBD. Additionally, the analysis was extended to differentiate between EBER-positive IBD and PIEBV+LPDs. The ROC curve analysis for these two diseases produced an AUC of 0.85, with a discriminative threshold established at 5.4×10² copies/mL with high specificity (95.65%) to ensure diagnostic accuracy for this rare but severe condition ( Figure 3C ).

The detection capacity achieved by plasma EBV DNA cutoff values indicating intestinal EBV infection was evaluated in the validation cohort. Analysis of clinical characteristics revealed no significant differences in demographic features and clinical presentations within the test and validation cohorts ( Supplementary Table S1 ). In the validation cohort, a threshold of 9.21×10¹ copies/mL for EBV DNA load in plasma demonstrated an LR⁺ calculated at 5.06, attaining values of 83.33%, 84.38%, 80%, and 87.1% in terms of specificity, sensitivity, NPV, and PPV, respectively, for identifying intestinal EBER-positive and EBER-negative diseases ( Table 2 ). For IBD, the sensitivity, specificity, PPV, NPV, and LR⁺ of a threshold of 5.28×10¹ copies/mL to differentiate intestinal EBV infection from non-EBV infection were 66.67%, 55.56%, 50%, 71.43%, and 1.5, respectively ( Table 2 ). Furthermore, at a cutoff value of 5.4×10² copies/mL, the sensitivity for distinguishing PIEBV+LPDs from EBER-positive IBD reached 85%, with an LR⁺ of 10.2 ( Table 2 ).

The analysis further explored and compared the six-month prognosis following intestinal EBER-ISH test among patients with intestinal diseases who were positive for EBV DNA load in plasma. The results confirmed that individuals in the intestinal EBER-positive group displayed significantly poorer prognoses than those in the EBER-negative group of the test cohort (50% vs. 15%, P < 0.001). Moreover, EBER-positive IBD and PIEBV+LPD patients demonstrated worse outcomes than their EBER-negative counterparts (26.1% vs. 15.2%, P =0.334 and 75% vs. 28.6%, P = 0.067, respectively). Details comparing the prognostic outcomes of various intestinal diseases are presented in Supplementary Table 2 .

Comparative analyses of plasma EBV DNA concentrations between the fatal and benign groups with various intestinal diseases were also conducted. As depicted in Figure 4A , patients with a fatal prognosis had greater median plasma EBV DNA loads than those with a benign outcome (2.9×10² copies/mL vs. 5.02×10¹ copies/mL, P = 0.0003). Furthermore, patients with PIEBV+LPDs and those with PINEBV+LPDs had significantly different median plasma EBV DNA levels (6.86×10² copies/mL vs. 5.09×10¹ copies/mL, P < 0.0001, Figure 4C ). However, patients with fatal prognosis did not show a statistically significant difference in plasma EBV DNA load compared to those with benign prognosis in IBD (P=0.8824) ( Figure 4B ).

ROC curve analysis revealed that there was an excellent connection between plasma EBV DNA concentrates and prognosis for intestinal EBV infection, demonstrating an AUC of 0.71 in the test cohort ( Figure 4D ), with a sensitivity of 69.7% and a specificity of 72%, demonstrating moderate ability to differentiate between fatal and benign prognoses. By establishing a threshold of 6.77×10¹ copies/mL for plasma EBV DNA load, we established the sensitivity, specificity, PPV, NPV, LR⁺, and LR^- for differentiating between fatal and benign prognoses in the validation cohort. The resulting values were 87.5%, 68.09%, 58.33%, 91.43%, 2.74, and 0.18, respectively ( Table 3 ). Furthermore, plasma EBV DNA concentration at 1.7×10³ copies/mL was shown to be the most effective threshold for predicting PILPD prognosis, with an AUC of 0.8 ( Figure 4E ), with a high specificity of 90.91% and relatively low sensitivity of 65%, ensuring the accurate identification of patients at risk of severe outcomes of PILPDs. Within the validation subset, the established cutoff offered a sensitivity of 64.71%, specificity of 83.33%, PPV of 91.67%, NPV of 45.45%, LR⁺ of 3.88, and LR^- of 0.42, effectively distinguishing between benign and fatal outcomes in PILPDs ( Table 3 ).