EP1081 - A Predictive Nomogram For Severe Fournier's Gangrene And The Imperative For Sequential SOFA Score Monitoring

Purpose/Background: Fournier's gangrene (FG) is a rare but potentially fatal soft tissue infectious disease primarily affecting the perianal and perineal regions. The early identification of severe FG remains a significant clinical challenge, with no established tools for risk stratification at presentation. To address this, we aimed to develop a nomogram using readily available clinical data that can be used to assess severe FG patients upon admission.

Methods/Interventions: A retrospective, single-center study of 107 FG patients was conducted from January 2013 to June 2024. We defined severe case as the presence of extra-perineal tissue involvement, organ dysfunction, or 28-day mortality. Based on this cohort, the 107 patients were randomly divided into a training set and a validation set at a ratio of 85% to 15%. Key predictors for the nomogram were identified by integrating the Least Absolute Shrinkage and Selection Operator (LASSO) regression with multivariate logistic analysis. The model's performance was thoroughly evaluated by the C-index, area under the receiver operating characteristic curve (AUC), calibration curves, and decision curve analysis (DCA), which was further confirmed in the validation set.

Results/Outcomes: A total of 107 patients were included, with 34 mild cases and 73 severe cases. The average surgical time, length of hospital stay, hospital charges, and ICU transfer rate in the severe group were greater than those in the mild group (P < 0.05). After feature screening, two key variables were selected to establish a nomogram: Neutrophil-to-Lymphocyte Ratio (NLR) and Sequential Organ Failure Assessment (SOFA). In the training set (n=90), it showed outstanding discrimination with a C-statistic of 0.996 (95% CI: 0.989-1). At an optimal threshold of 0.588, sensitivity was 98.4% and specificity was 96.2%. These results were confirmed in the validation set (n=17). The model also exhibited good calibration (Hosmer-Lemeshow test P=0.948) and provided clinical net benefit across a threshold probability range of 0-40%, as shown by decision curve analysis. In addition, an intragroup analysis was conducted on the severe group. In contrast, we found that the proportion of widespread infections in the severe death group was lower (37.93% vs. 6.67%, P=0.044). The SOFA score continued to increase on the seventh day after surgery (3.0 vs. 9.0, P< 0.001).

Conclusion/Discussion: Combining the NLR with the SOFA score can be used to construct a nomogram for effectively identifying severe FG patients. The SOFA score should be continuously monitored during treatment.

Figure 1. Flowchart for patient selection
Figure 2. The model discriminative and calibration performance
(A)AUC curves of the models: (A-1) Development set; (A-2) Validation set. (B) Calibration curves of the models: (B-1) Development set; (B-2) Validation set. (C) DCA of the model. (D) Nomogram for identifing mild and severe FG patients. (E) The nomogram was applied to predict disease severity in a patient with FG: a 63-year-old male with an NLR of 3.59 and a SOFA score of 3. After plotting these values and summing the corresponding points, the total score reached approximately 31, which corresponded to a predicted probability of severe disease of 0.964. This high probability would alert clinicians to prioritize intensive care unit admission and initiate an aggressive treatment protocol.