Introduction

Hepatic echinococcosis, a global zoonotic parasitic disease, is caused by cestodes of the genus Echinococcus (family Taeniidae)1. Its two primary human forms are cystic echinococcosis (caused by Echinococcus granulosus) and alveolar echinococcosis (caused by Echinococcus multilocularis)2. Surgical techniques are pivotal in treating hepatic echinococcus2.

Postoperative complications, especially pulmonary ones such as infection and pleural effusion are common. Postoperative pulmonary complications (PPCs) discomfort the patients, prolong their hospital stays, and elevate healthcare costs3. Enhanced recovery after surgery (ERAS) principles, including perioperative airway management, have been developed to address these issues4. Understanding PPCs risk factors is vital for improving perioperative airway management efficacy. However, studies explicitly examining PPCs following surgical interventions for hepatic echinococcosis are limited and frequently constrained by small sample sizes5,6. Therefore, this study retrospectively analyzed case records from patients who were surgically intervened for hepatic echinococcosis at Qinghai Provincial People’s Hospital. It was aimed at identifying the predisposing factors associated with PPCs. A prediction model was developed utilizing these identified factors for PPCs in hepatic echinococcosis, offering valuable insights for future prevention and treatment strategies.

Methods

Research participants

Data of patients treated surgically for hepatic echinococcosis from January 2016 to June 2023 was collected from Qinghai Provincial People’s Hospital. We confirmed that all methods were performed in accordance with the relevant guidelines and regulations. This experiment, exempted from informed consent, was approved by the Ethics Committee of Qinghai Provincial People’s Hospital, China (2023 − 281).

Inclusion criteria

Patients diagnosed and surgically treated for hepatic echinococcosis.

Exclusion criteria

(1) Patients with echinococcosis in other sites. (2) Patients who combined operations, excluding cholecystectomy. (3) Patients requiring reoperation for complications from previous operations. (4) Patients who underwent liver transplantation. (5) Patients who died within 24 h postoperatively. (6) Patients with malignant tumors. (7) Patients with incomplete case data.

Diagnostic criteria for PPCs

PPCs are respiratory conditions occurring within 30 postoperative days. Currently, no universal PPCs diagnostic criteria exist. In this study, the European perioperative clinical outcome (EPCO) definitions for PPCs were primarily utilized and further enhanced by our clinical expertise (Table 1)7. PPCs adhered to the Clavien–Dindo standards8.

Table 1 Diagnostic criteria of postoperative pulmonary complications.

Surgical methods

Surgical approach selection for hepatic echinococcosis should follow the radical treatment principle2. However, due to patient-specific variations, surgical choices should be personalized. For alveolar echinococcosis, besides liver transplantation, options include curative resection, palliative lesion resection, necrotic liquified cavity puncture drainage, and local ablation. Standard procedures for cystic echinococcosis encompass enucleation of the endocyst, total cystectomy, sub-total cystectomy, partial hepatectomy, and percutaneous “PAIR” (puncture, aspiration, injection, and re-aspiration)2.

These methods were categorized into four operation types and three modes for statistical purposes. The operation types are laparotomy, laparoscopic operation, local ablation, and puncture drainage (including necrotic liquified cavity puncture drainage and percutaneous “PAIR”). When two operations were performed simultaneously, the more complex type was documented during data collection. For instance, laparoscopic segment III hepatectomy with local ablation was recorded as a laparoscopic operation. Non-hepatectomy procedures include enucleation of endocyst, total or sub-total cystectomy, puncture drainage, and local ablation; other surgical procedures were classified as hepatectomy. The liver was divided into eight segments based on the portal and hepatic vein distribution9, with hepatectomies classified as major (three or more segments resected) or minor (less than three segments resected).

Data collection

To address the lack of studies on risk factors for PPCs in hepatic echinococcosis surgery, we reviewed and analyzed existing literature on PPCs following hepatectomy5,6,10. Based on these references and our experience, the research questionnaire was constructed as follows:

General information

Inpatient number, sex, age, body mass index (BMI), hospital stay duration.

Medical history

Smoking and drinking history, cardiovascular conditions (including congenital heart disease, coronary atherosclerotic heart disease, and arrhythmia), diabetes, hypertension, cerebrovascular accidents, liver disease (hepatitis, cirrhosis), and lung disease (such as chronic obstructive pulmonary disease, asthma, and tuberculosis).

Preoperative information

Disease characteristics (number of foci, focal diameter, hydatid pathology types), preoperative antibiotic use within 1 month, hypoproteinemia (< 35 g/L), anemia (male: <120 g/L, female: <110 g/L), and total bilirubin concentration.

Intraoperative information

Operation type and mode, blood loss volume, the need for blood transfusion, and operation duration.

Statistical analysis

Continuous variables were expressed as mean ± standard deviation (SD), and categorical variables as percentages. Chi-square tests or Fisher exact tests were sued to assess disparities in categorical variables. Independent sample t-tests or Mann–Whitney U tests analyzed continuous variables. Statistical significance was set at P < 0.05. Analyses were performed using SPSS 25.0 software and R language 4.3.1 software.

Sample size calculation

A 60% PPCs incidence rate was assumed for patients with hepatic echinococcosis. And a C-index assumption of 0.8 was used to calculate the sample size. The number of variable parameters was proposed to be 23. Utilizing the “pmsampsize” package in R, we calculated that the training set would consist of 696 patients. Calculate the sample size required to construct a nomogram model for PPCs with Clavien-Dindo Grade ≥ II using the same method.

Construction and validation of the prediction model

Using the “car” and “survival” packages in R, we randomly divided the enrolled patients into training and validation sets. The training set was utilized to develop a nomogram prediction model, and the validation was used to assess its performance. Least absolute shrinkage and selection operator (LASSO) regression analysis identified risk factors for PPCs, and multivariate logistic regression analysis determined the variables for the final model. Subsequently, a nomogram prediction model (model 1) was constructed to predict PPCs. To evaluate the model 1’s performance, metrics including ROC, calibration plots, Hosmer-Lemeshow test, and decision curve analysis (DCA) were employed. Furthermore, in order to enhance the clinical guidance significance, we will consider PPCs with Clavien-Dindo Grade ≥ II as positive events and develop a new nomogram model (model 2). In addition, the reliability and discriminative ability of the model 2 will be evaluated.

Quality control

The research plan was determined after the examination and approval by the hepatic echinococcosis experts. Before data collection, researchers were trained on the diagnostic criteria, case inclusion and exclusion criteria of PPCs. In the process of data collection, the inclusion and exclusion criteria were strictly followed, and the data entry was jointly operated by two researchers to check and maintain the accuracy, consistency and integrity of the data.

Results

Incidence of PPCs

Data were collected from 1657 patients. According to the exclusion criteria, 480 were excluded: 323 with echinococcosis in other sites, 121 with combined operations (excluding cholecystectomy), 7 needing reoperation due to complications from previous operations, 12 having undergone liver transplantation, 8 dying within 24 h post - operation, 6 with malignant tumors, and 3 due to incomplete medical records. Finally, 1177 patients who met all requirements were retained. Among the 1177 patients, 690 experienced the development of PPCs, resulting in an incidence rate of 58.6%. Table 2 details the incidence rates of diverse types of PPCs. Significantly, Clavien-Dindo Grade I represented approximately 40% (475 of 1177), and the incidence rate of PPCs (Clavien-Dindo Grade ≥ II) was only 18.3%, as shown in (Table 3).

Table 2 Incidence of postoperative pulmonary complications.
Table 3 Clavien-Dindo classification of postoperative pulmonary complications.

Comparison of clinical characteristics between PPCs and non-PPCs groups among patients who underwent surgery for hepatic echinococcosis

Patients who underwent surgery for hepatic echinococcosis were categorized into two groups based on PPCs occurrence: those with and without PPCs. Significant differences were observed between the two groups in the following parameters: length of hospital stay, age, smoking and drinking history, pre-existing lung disease, preoperative antibiotic use within 1 month, hypoproteinemia, anemia, number and diameter of foci, hydatid pathology characteristics, type and mode of operation, blood loss volume, the need for blood transfusion, and operation duration (SDC Table 1).

Comparison of clinical features among patients who underwent surgery for hepatic echinococcosis in the training and validation sets

Following a 7:3 ratio, 824 patients were randomly assigned to the training set and 353 to the validation set. No significant differences in any factors were observed between the sets (P > 0.05) (SDC Table 2).

Identification of PPCs predictors

PPCs occurrence was the dependent variable, and the remaining research variables except length of stay were served as independent variables. LASSO regression analysis on the training set data revealed high-risk factors for PPCs. Totally 9 high-risk factors were identified at lambda.1se: body mass index (BMI) ≥ 28 kg/m2, smoking, pre-existing lung disease, focal diameter > 10 cm, major hepatectomy, antibiotic therapy within 1 month before surgery, blood loss volume ≥ 400 mL, the need for blood transfusion, and operation duration ≥ 3 h.

Subsequent multivariate logistic regression analysis confirmed 7 factors as independent risk factors for PPCs in patients with hepatic echinococcosis, including BMI, pre-existing lung disease, focal diameter, mode of operation, antibiotic therapy within 1 month before surgery, the need for blood transfusion, and operation duration (SDC Table 3).

Construction and validation of the model 1 for predicting PPCs in hepatic echinococcosis

Model 1 was constructed by utilizing 7 independent risk factors (Fig. 1). Its area under the receiver operating characteristic curve (AUC) value in the training set was 0.808 (95% confidence interval [CI], 0.779–0.838), with a probability threshold of 0.612 (coordinates: x = 0.786, y = 0.697) (Fig. 2a). The AUC for the model applied to the validation set was 0.791 (95% CI, 0.744–0.838), with a probability threshold of 0.629 (coordinates: x = 0.854, y = 0.658) (Fig. 2b). This model revealed superior discriminatory ability in predicting PPCs among patients with hepatic echinococcosis.

Fig. 1
figure 1

The nomogram constructed for predicting PPCs in hepatic echinococcosis. BMI body mass index, PPCs postoperative pulmonary complications.

Fig. 2
figure 2

Receiver operating characteristic (ROC) curves for the prediction model ((a) training set; (b) validation set).

The calibration plots for both sets (Fig. 3a, b) indicated a strong concordance among the apparent, bias-corrected, and ideal lines. The Hosmer–Lemeshow test indicated a P-value of 0.691, suggesting an accurate prediction of actual versus predicted probabilities. The DCA of the training set revealed that clinical implementation of this predictive model yields net benefit across a wide threshold probability range (18-99%), indicating its potential to enhance patient outcomes when guiding therapeutic interventions. (Fig. 4a). Similar outcomes were observed in DCA performed on the validation set (Fig. 4b).

Fig. 3
figure 3

Calibration plots for the model ((a) training set; (b) validation set).

Fig. 4
figure 4

DCA for the model ((a) training set; (b) validation set).

Construction and validation of the model 2 for predicting PPCs of Clavien-Dindo Grade ≥ II in hepatic echinococcosis

The construction of model 2 for PPCs with Clavien-Dindo Grade ≥ II required a sample size of 1,020; therefore, data from all 1177 patients were included in the model development. And then the performance of the model 2 was assessed. LASSO regression analysis identified 3 high-risk factors: mode of operation, the need for blood transfusion and operation duration. Subsequently, through multivariate logistic regression analysis, all 3 high-risk factors were identified as independent risk factors.

Next, we developed the model 2 founded on the 3 identified independent risk factors (Fig. 5). Its AUC value in the training set was 0.736 (95% CI, 0.702–0.771), with a probability threshold of 0.186 (coordinates: x = 0.663, y = 0.707) (Fig. 6a).

Fig. 5
figure 5

The nomogram constructed for predicting PPCs of Clavien-Dindo Grade ≥ II in hepatic echinococcosis.

Fig. 6
figure 6

Evaluation results of PPCs (Clavien-Dindo Grade ≥ II) nomogram model ((a) ROC; (b) calibration plots; (c) DCA).

The Hosmer–Lemeshow test revealed a P-value of 0.99. The calibration plots also indicated a good consistency among the apparent, bias-corrected, and ideal lines (Fig. 6b). The DCA demonstrated that when the threshold probability ranges from 5 to 38%, utilizing this model to guide treatment decisions may improve clinical outcomes. (Fig. 6c).

Discussion

This study contributes valuable insights into the incidence rates and associated risk factors related to PPCs in hepatic echinococcosis. In our study, the predominant treatment method was hepatectomy, which was performed in 59.2% of the 1177 patients. Other surgical procedures included enucleation of endocyst, total or sub-total cystectomy, puncture drainage, and local ablation. Although each specific surgical procedure varied, all interventions had implications for diaphragm function, leading to potential complications such as pleural effusion, atelectasis, hypoxia, and subsequent respiratory impairment11,12. This impact results in a significant incidence of PPCs. However, the existing literature lacks a universal definition for PPCs. The widely referenced EPCO criteria encompass 7 common pulmonary symptoms, but the clinical practice reveals a broader spectrum of PPCs. The choice of PPCs definitions can influence the identification of various risk factors13,14,15. This study expanded the conventional scope by including conditions such as pleurisy, pyothorax, pulmonary embolism, emphysema, pulmonary edema, and acute respiratory distress syndrome.

In our study, the PPCs incidence rate was 58.6%. Previous studies reported that it ranged from 2 − 70% after hepatectomy16. Variances in PPCs incidence across studies may stem from two main factors. First, different definitions of PPCs are used. For example, if PPCs were defined as Clavien-Dindo Grade ≥ II, the incidence rate was only 18.3%. Second, the included surgical procedures matter. When the study included only hepatic echinococcosis patients who underwent hepatectomy, the PPCs incidence rate was 71.3%.

PPCs significantly prolonged hospital stays to an average of 27.22 ± 12.53 days in our study. Therefore, identifying risk factors for PPCs and intervening early is crucial. Although the comparison of clinical characteristics between the PPCs group and the non-PPCs group revealed significant differences in 15 factors, only 7 factors were identified as independent factors by LASSO and Logistic regression analyses. If PPCs were defined as Clavien-Dindo Grade ≥ II, just 3 factors were determined to be independent factors. We demonstrated that mode of operation, the need for blood transfusion and operation duration had a significant impact on the occurrence of PPCs.

In this study, the incidence of PPCs increased significantly with the rising complexity of surgical methods. The incidences of PPCs in non-hepatectomy patients, minor hepatectomy patients, and major hepatectomy patients were 44.63%, 56.54%, and 81.40%, respectively. Major hepatectomy involves intricate surgical procedures, including repeated occlusion of blood flow into the liver and prolonged ischemia-reperfusion injury, which induce the release of inflammatory cytokines. Cytokines can trigger a systemic inflammatory response and affect respiratory mechanics by decreasing lung tissue elasticity and compliance17, thereby promoting PPCs. Although the assessment methods for the extent of liver resection vary, our study still reached the same conclusion as the previous one16: liver resection is an independent risk factor for PPCs and has a vital impact on the severity of PPCs.

A study revealed a prominent association between blood transfusion and the incidence of PPCs18, which is consistent with our findings. Blood transfusion, particularly the infusion of red blood cells, can exhibit more substantial adhesion to endothelial cells and accumulate them in the lungs, reducing blood oxygen and promoting microangiopathopathy and inflammation19. Moreover, transfusion-related immunomodulation can weaken cell-mediated immunity by suppressing the activity of natural killer cells and cytotoxic T cells20, raising the susceptibility of PPCs in patients receiving transfusions.

Furthermore, we confirmed a considerable association between surgical duration > 3 h and an increased risk of PPCs, which is in accordance with the study conducted by Smetana GW21. Extended operation time implies prolonged exposure to anesthesia, and it elevates the risks inherent in the surgical procedure. In this study, general anesthesia, which inhibits the respiratory center’s regulation, was universally performed. Routine mechanical ventilation under general anesthesia can cause hypoxemia in lung tissues and damage respiratory function22. Therefore, optimizing ventilation techniques during general anesthesia is imperative. Pulmonary protective measures, such as low tidal volume and minimal positive end-expiratory pressure (PEEP), can effectively lower the risk of developing PPCs23. While the ideal tidal volume and PEEP settings are still under review, evidence suggests that 6–10 mL/kg for tidal volume and 2–5 cm H2O for PEEP are appropriate24.

A previous study indicated a remarkably higher incidence of PPCs in obese and overweight patients post-hepatectomy compared to those of average weight25. Coincide with these findings, our study showed a 0.364-fold increase in PPCs incidence among patients with BMI ≥ 28 kg/m2. This susceptibility may involve physiological mechanisms. Specifically, abdominal obesity causes an accumulation of adipose tissue in both the abdomen and the chest, which intensifies pressure on the lungs by pushing the diaphragm upward. Consequently, this generates reduced lung capacity, increased pleural pressure, decreased chest wall compliance, and diminished functional residual capacity24,26. Different from these researches25, obesity is not an independent risk factor for PPCs > Grade I in our study. This may be related to the inconsistent definitions of overweight and obesity, as well as differences in surgical methods and complication types.

Pre-existing lung conditions significantly increase the risk of PPCs27, possibly due to chronic airway inflammation, excessive secretions, structural damage to lung tissue, or immunosuppression from prolonged preoperative intensive use of steroids. It is advisable for patients with such respiratory diseases to manage their conditions effectively before elective surgeries and maintain appropriate pulmonary treatments through the perioperative period to minimize the risk of PPCs28,29. In this study, pre-existing lung conditions were identified as an independent risk factor for PPCs, but not for those with Clavien-Dindo Grade II or above. It is speculated that pre-existing lung conditions have an important influence on the occurrence of PPCs, but have little effect on their severity. The same influence on PPCs was observed in focal diameter and antibiotic therapy within 1 month prior to surgery.

A focal diameter exceeding 10 cm impacts the patients’ nutritional status negatively30, leading to a diminished ventilatory drive, weakened respiratory muscle function, lung parenchymal changes, and compromised lung defenses31, thereby increasing the susceptibility to PPCs32. Moreover, if the focal diameter exceeds 10 cm, the extent of liver involvement can intensify33, leading to a significant reduction in liver reserve capacity and diminished surgical tolerance, thereby elevating the risk of postoperative liver failure34. The correlation between PPCs post-hepatic resection and the incidence of postoperative liver failure has been established35. Furthermore, as hydatid lesions enlarge, they increase abdominal pressure and impair diaphragmatic functionality, predisposing the patients to potential preoperative respiratory involvement and increasing the likelihood of developing PPCs. Although we identified this factor as inalterable, if early screening can be performed and timely treatment can be taken before the lesion surpasses 10 cm, then the incidence of postoperative complications may be lower.

When collecting cases, we noticed that although some patients had no respiratory disease prior to surgery, they had a history of treatment for respiratory disease within 1 month before an operation. Therefore, we analyzed the use of antibacterial drugs within 1 month before surgery as one of the risk factors, and the results showed that it was one of the independent risk factors for the occurrence of PPCs and there was a positive correlation. The possible explanation is that although these patients have no clinical manifestations in the preoperative period, they maintain their susceptibility to respiratory diseases.

The model 1 demonstrated excellent discriminative ability, consistency, and clinical relevance, as validated by ROC, calibration plots, Hosmer-Lemeshow test, and DCA. Furthermore, considering that the prevention and treatment of Clavien-Dindo Grade ≥ II PPCs may be more concerned in clinical work, we further established a model 2 using this criterion as positive events and the AUC was 0.736. However, the model 2 demonstrated seemingly limited predictive capacity, with a maximum predicted risk of only 0.3 for Clavien-Dindo Grade ≥ II PPCs. This constrained performance likely reflects the condition’s inherently low baseline incidence, though methodological limitations including suboptimal feature selection and class imbalance may also contribute. In clinical practice, the two models can be implemented in a combined approach. Clinical decision pathway: PPCs risk < 18%: No therapeutic intervention required. PPCs risk 18–61.2%: Implement close monitoring and dynamic reassessment. PPCs risk > 61.2%: Initiate general treatments (e.g., chest physiotherapy, mucolytic agents). Concurrently apply model 2: If the calculated disease risk ≥ 18.6%, administer aggressive therapies (e.g., antibiotics, drainage procedures). Through this application framework, clinicians can stratify target patients to achieve both active disease management and avoidance of overtreatment.

In our study, an accessible and straightforward prediction model for PPCs among patients operated on for hepatic echinococcosis aims to help healthcare providers identify these patients early and provide individualized therapy; however, our study has its limitations. Operative procedures for hepatic echinococcosis are typically performed under general anesthesia, with mechanical ventilation being a significant risk factor for PPCs. Over 25% of patients with healthy lungs may suffer various lung injury from mechanical ventilation22. With ERAS protocols adopted in China, lung-protective ventilation strategies have been employed during operations for some patients. However, the lack of detailed data on life support during anesthesia prevented this study from categorizing these ventilation strategies and assessing the effect of these techniques on the occurrence of PPCs among hepatic echinococcosis patients. The majority of included patients were from the Qinghai-Tibet Plateau. Reports indicate that characteristics of hydatid disease vary between domestic and international settings36, as do surgical techniques across different regions, potentially affecting this model’s widespread applicability. Therefore, further larger-scale studies are warranted.