PDF
Citation
Print
INTRODUCTION
The prevalence of gallbladder (GB) polyps in adults was reported to be 0.3 to 9.96% and is expected to increase with an aging society and more accessible health screening.1,2 Although most GB polyps are discovered incidentally on abdominal ultrasonography (US) or computed tomography (CT) scans and do not require treatment, neoplastic lesions should be treated because they can progress to GB cancer.3,4 Laparoscopic cholecystectomy is a highly effective treatment for GB polyps that can lead to less pain, shorter hospital stays, and a better quality of life. On the other hand, there is a risk of post-cholecystectomy syndrome, which can cause various gastrointestinal symptoms, such as indigestion, abdominal pain, or steatorrhea.5 Furthermore, the post-cholecystectomy state may increase the risk of metabolic diseases such as cardiovascular disease and diabetes.6-8 Therefore, individualized decision-making for each patient is essential to reduce the incidence of unnecessary cholecystectomies.
Clinical guidelines have proposed treatment algorithms for GB polyps based on the clinical factors.9-11 On the other hand, cholecystectomy was performed in several cases according to guidelines out of concern for a neoplastic polyp, but the pathology specimen confirmed a non-neoplastic condition that did not require treatment.12 Furthermore, most guidelines do not specify the roles and limitations of CT and US in predicting GB neoplastic polyps. Several studies have proposed advanced diagnostic modalities, such as artificial intelligence or endoscopic ultrasound, for a better preoperative prediction of neoplastic polyps.12-15 These innovative techniques are still under development and have limitations in terms of accessibility.
This study assessed the real-world utility of seven surgical indications for cholecystectomy, as outlined in the clinical practice guidelines, in predicting neoplastic GB polyps. In addition, this study aimed to identify candidate factors that could enhance the preoperative prediction of neoplastic lesions.
SUBJECTS AND METHODS
1. Ethical approval and informed consent
This study was approved by the Institutional Review Board of the authors' institution (June 15, 2023, approval number VC23RISI0135) and was conducted following the Declaration of Helsinki. The participants' data were anonymized before analysis to ensure privacy and confidentiality. Informed consent was not required because of the nature of the study.
2. Study design
This retrospective study recruited patients who underwent cholecystectomy for GB polyps at the authors' institution between May 2009 and December 2019. Those with identified surgical indications through medical record review were included in the study population. Patients with insufficient clinical or preoperative imaging data were excluded. Fig. 1 presents the recruitment and dropout process of the study population.
3. Data collection
The patients’ demographic information, including age, sex, weight, height, smoking and drinking habits, and accompanying medical history, was collected. The preoperative laboratory findings, including tumor markers, e.g., carcinoembryonic antigen (CEA) and CA 19-9, were also collected. The indications for cholecystectomy for GB polyps were classified into the following seven categories by referring to the literature: 1) polyp size >10 mm, 2) increase in polyp size during follow-up, 3) symptomatic patients, 4) patients aged ≥50 years, 5) concomitant gallstones, 6) a single lesion, and 7) a sessile morphology.9-11 Multiple surgical indications for each patient were reflected separately in each corresponding item. The preoperative imaging was interpreted by a single radiologist who reviewed the preoperative CT images of the GB to assess the number, size, shape, and associated findings of GB polyps. Three clinicians with more than two years of clinical experience in pancreaticobiliary diseases reviewed the preoperative US images. One clinician investigated the findings on the examiner's reports, while the other clinician reinterpreted all preoperative images for cross-validation. The three clinicians resolved discrepancies between the examiner's reports by conducting an additional image review and discussion before determining the final findings based on unanimous consensus. The presence of gallstones or GB wall thickening (defined as cases in which the measured wall thickness was >4 mm16,17) in the preoperative images was noted. The pathology reports of surgical specimens were also reviewed. The study group was divided into two groups based on the pathology reports: neoplastic polyps (polyps containing an adenomatous component or cancer cells) or non-neoplastic polyps (including cases with no polyp on the surgical specimen). This study compared the differences in demographics, medical history, laboratory findings, surgical indications, preoperative imaging, and pathological characteristics of patients with and without neoplastic polyps.
4. Preoperative prediction for GB neoplastic polyp
A receiver operating characteristic (ROC) curve was plotted for each surgical indication. The area under the curve (AUC), sensitivity, specificity, and positive and negative predictive values for each surgical indication were calculated to compare the predictive performance for neoplastic polyps. The same approach was used for the polyp size alone in US and CT separately. The cutoff value for the polyp size in the US and CT images for the optimal AUC value was also obtained. Logistic regression analysis was also conducted on the clinical variables and surgical indications to ascertain the factors associated with neoplastic polyps.
5. Statistical analysis
The continuous variables are presented as the mean and standard deviation (SD), and categorical variables are presented as frequency and percentage. The differences among variables were analyzed using a Student's t-test or Wilcoxon– Mann–Whitney tests for the continuous variables and Chi-square or Fisher's exact tests for the categorical variables. Univariate logistic regression analysis was performed for each candidate variable, with pathological neoplastic polyps as the outcome variable, to identify the candidate predictors of neoplastic polyps. Subsequent multivariate analysis was performed on the variables with p<0.2 in univariate analysis. The final regression model was fitted using stepwise backward regression based on the multivariable analysis results. The R-4.2.2 program (The R Foundation for Statistical Computing, r-project.org) was used for all statistical analyses and data visualization.
RESULTS
1. Baseline characteristics
The data for 385 patients were analyzed (mean age, 49.9 years; 55.6% males). Except for two patients, the study group underwent a laparoscopic cholecystectomy. The mean CEA and CA 19-9 levels before the cholecystectomy were within the normal ranges (Table 1). Among the surgical indications, cholecystectomy for polyp sizes larger than 10 mm was most common (78.2%), followed by a single lesion, concomitant gallstones, and symptomatic patients (Supplementary Table 1).
Table 2 lists the results of preoperative imaging studies and the pathological findings of surgical specimens. The mean size of the measurable polyps was larger in the US images than in the CT images. Three hundred and twenty- eight patients (85.2%) had actual polyps in their surgical specimens; more than half (58.5%, n=192) had two or more polyps. Cases with more than four polyps or concomitant gallstones were observed more frequently in pathological specimens compared to the preoperative images of the same patient. Concomitant gallstones were observed more frequently in pathological specimens than in preoperative imaging (32.7% vs. 18.4%, p<0.001).
The cholesterol polyp (n=231, 70.4%) was the most prevalent type of GB polyp observed. Neoplastic polyps were identified in 62 patients (18.9%), with 13 of them being diagnosed with GB cancer.
2. Difference between neoplastic and non-neoplastic polyps
Patients with neoplastic polyps in the cholecystectomy specimens were older and comprised more females than those with non-neoplastic polyps. No significant differences in body mass index, concomitant diseases, alcohol consumption, or smoking habits were observed between the two groups. The preoperative laboratory findings showed that the neoplastic polyp group had lower platelet counts than the non-neoplastic polyp group. The neoplastic polyp group exhibited a significantly higher rate of polyps >10 mm and a larger mean polyp size measured on CT and US (p<0.001). The pathological results showed that the neoplastic polyp group had larger polyps and a higher proportion of solitary lesions than those in the non-neoplastic polyp group. On the other hand, concomitant gallstones were observed more commonly in the non-neoplastic polyp group (Table 3).
3. Surgical indications and GB neoplastic polyps
Fig. 2 shows the ROC curves and predictive performances of the seven surgical indications for predicting neoplastic polyps. The AUC values for individual surgical indications were unsatisfactory, ranging from 0.5 to 0.6. The surgical indication with the highest AUC value was 'polyp size >10 mm' in the preoperative images (AUC 0.601, sensitivity 95.2%, specificity 25.1%). The three surgical indications—'polyp size ≥10 mm,' 'increase in polyp size,' and 'patient's age ≥50 years'—showed a high sensitivity of over 90% despite the poor specificity. In contrast, the surgical indications, 'symptomatic patients' and 'a sessile morphology,' showed reasonable specificities of 87.0% and 90.1%, respectively, while their sensitivity remained very low.
4. GB polyp sizes in preoperative images and surgical specimens
In the surgical specimens, GB polyps measuring 10 mm or more showed a significantly higher proportion of neoplastic lesions (77.4% vs. 26.0%, p<0.05). Regarding the preoperative imaging findings, Fig. 3 shows the predictive performances of neoplastic polyps based on the polyp size, as observed on both abdominal CT and US. The polyp size in the abdominal CT images showed a slightly higher AUC value than in the US images (CT: 0.798; US: 0.765). The optimal cutoff value was set higher in US than in CT (12 mm vs. 10 mm). At this cutoff, both methods exhibited similar diagnostic performances. When using a 10 mm cutoff size for polyps in the US images, as commonly recommended in the clinical guidelines, the sensitivity for predicting neoplastic lesions was superior to CT (93.5% vs. 66.1%). On the other hand, specificity and positive predictive value (PPV) were significantly lower than with a 12 mm cutoff size (specificity: 26.6%; PPV: 19.7%).
The proportion of pathologic neoplastic polyps was 30.3% (47 out of 155) when a 12 mm cutoff size was applied to the US images (Table 4). When combined with a 10 mm cutoff size for polyp size on CT, the frequency of pathological neoplastic polyps was significantly higher in those who had polyps ≥10 mm on CT (46.1% vs. 15.2%, p<0.05), suggesting that combining US and CT size measurements can reduce the number of cases of cholecystectomy for non-neoplastic polyps compared to making decisions based solely on the size from US images.
5. Clinical variables related to predicting neoplastic polyps
Among seven surgical indications, the patient's age and polyp size were used as variables. Separate logistic regression analyses were conducted based on the polyp size from either CT or US (Tables 5 and 6). Based on univariate analysis, the following variables were selected for multivariate analysis: polyp size, sessile morphology, age, sex, platelet count, and gamma- glutamyl transferase. After stepwise selection, the final multivariable regression model included four variables: polyp size on the CT or US images, age, sex, and platelet count. Larger polyps, older age, female sex, and lower preoperative platelet count were associated with a higher likelihood of pathological neoplastic polyps. The results were consistent regardless of the measurement modality used for the polyp size (i.e., CT or US).
DISCUSSION
This study examined the indicators for predicting neoplastic polyps before cholecystectomy in patients with GB polyps. Among the 385 patients who underwent cholecystectomy for GB polyps, more than 80% of patients had non-neoplastic polyps or no polypoid lesions, suggesting that they did not require surgery. Hence, the current surgical indications of cholecystectomy for GB polyps have an unsatisfactory ability to predict the presence of neoplastic polyps. The polyp size in the US and CT images was the most influential variable in predicting neoplastic polyps. A more selective surgical decision could be made for neoplastic polyps by combining the findings of CT and US rather than relying on a single imaging modality. In addition to the polyp size, the patient's age, sex, and preoperative platelet count were also clinical variables associated with neoplastic GB polyps.
Adenomatous polyps are representative GB neoplastic polyps, accounting for approximately 5–10% of all GB polyps.18,19 Adenomatous polyps have the risk of progression to adenocarcinoma, and adenoma-to-carcinoma conversion is considered a crucial mechanism of carcinogenesis.20 The proportion of adenomatous polyps identified in this study was 16.2% (n=62), which was higher than that reported in previous studies, possibly because the study group included patients who had undergone cholecystectomy only for GB polyps.
The GB polyp size measured using US or CT had the greatest influence in predicting neoplastic polyps. Many clinical guidelines have shown that a preoperative polyp size larger than 10 mm is the most crucial clue for treatment decision- making for GB polyps.9-11 In the present analysis, the optimal cutoff value for the GB polyp size on CT for neoplastic polyp discrimination was 10 mm, which is consistent with existing guidelines and with favorable sensitivity, specificity, and AUC values. Nevertheless, differing opinions exist on whether the 1 cm polyp size criterion is insufficient for diagnosing neoplastic polyps.18,21 In a large-scale cohort analysis conducted in the Netherlands, the indication for surgery with a polyp size of 1 cm had moderate diagnostic accuracy. Moreover, approximately one-third of patients received unnecessary treatment, assuming that cholecystectomy had been performed based on the size criterion alone.21 In addition, considering the varying cutoff values for polyp size suggested by several studies,22,23 the heterogeneity of the study groups or different applications of measuring modalities may account for the non-concordance of the proposed cutoffs. Similarly, the data showed that the optimal cutoff polyp size for neoplastic polyps differed between CT and US: CT showed a smaller cutoff polyp size than US (10 mm vs. 12 mm). Further research will be needed to establish the optimal preoperative imaging criteria for cholecystectomy decisions based on the GB polyp size.
Abdominal US is a commonly used imaging modality for the initial detection of GB polyps, and a cholecystectomy is often decided based on the US findings at the time of diagnosis. Abdominal CT is also used for a preoperative evaluation when it is accessible, depending on the region and country.24 Compared to CT, US has high accessibility, real-time imaging, and no radiation risk. US is a useful screening tool for GB polyps, but its reported sensitivity and false positivity rates are approximately 36–90% and 6–43%, respectively.25 Considering that the examiner's subjectivity can influence US, the size of the lesions detected during image acquisition may be overestimated. In addition, US has limitations, including the potential for interference from artifacts such as bowel gas or excessive visceral fat. In decision-making, relying solely on the polyp size obtained from US images could lead to inappropriate treatment.
The proportion of neoplastic polyps was remarkably higher in patients who met the 10 mm cutoff on CT and the 12 mm cutoff on US compared to the other cases. Hence, combining these two imaging techniques can reduce unnecessary treatments and prevent missing polyps that should be resected. CT helps evaluate GB polyps larger than 5 mm and can determine the presence of hepatic invasion, regional lymph nodes, or distant metastasis.26 CT can also help predict neoplastic polyps by revealing polyps on plain images or showing indistinct margins or hyperenhancement on contrast- enhanced images.27 Considering the characteristics of each imaging modality, improved decision-making for GB polyps may be achieved using these two modalities concurrently during preoperative examinations because they complement each other. Future research should determine how CT and US can best complement each other in diagnosing neoplastic GB polyps.
Older age, female sex, and lower platelet count were also associated with neoplastic polyps according to multivariable logistic regression analysis. A positive correlation between the risk of neoplastic polyps and increasing age was confirmed, likely reflecting the average age of patients diagnosed with GB polyps (60–70 years old).28 Females had a higher risk than males, consistent with results from a previous report.29 Several studies have reported a potential relationship between solid organ malignancies (including GB cancer) and platelet count. On the other hand, most of these studies focused on thrombocytosis rather than decreased platelet count.30 The data showed a negative correlation between neoplastic polyps and the platelet count. Nevertheless, considering the low odds ratio in logistic regression analysis, further studies with more robust data will be needed to validate the association between thrombocytopenia and GB neoplastic polyps.
In contrast, coexisting gallstones were observed more frequently in pathologic reports of the non-neoplastic polyp group, with 114 patients (35.3%). The correlation between gallstones and GB polyps is still controversial. In the present data, among these 114 patients, cholesterol polyps were most common, accounting for 76 cases, suggesting that cholesterol metabolism plays a role in the shared mechanism underlying the formation of cholesterol polyps and gallstones. In particular, in 35 patients (30.7%), no gallbladder polyps were identified in the surgical specimens, suggesting that gallstones may be misinterpreted as polyps on preoperative imaging. This underscores the need for careful differentiation between polypoid lesions and gallstones during preoperative evaluation to avoid unnecessary surgeries.
This study had several limitations. First, access to more detailed information on the relevant variables of the neoplastic polyps, such as comorbidities, smoking, and alcohol consumption, was limited owing to the retrospective design. Second, potential bias may have occurred based on region, country, and ethnicity because the study was conducted at a single institution in Asia, making it difficult to generalize the findings. Multi-institutional prospective approaches are required for subsequent research. Third, verification of the level of imaging equipment and observers might be insufficient because this study analyzed preoperative US or CT images obtained by other institutions. Potential confounding factors might influence the interpretation of variables, such as polyp size and GB wall thickening, because interobserver bias is more likely in US than in CT. An attempt was made to compensate by having all imaging data of the study participants reviewed again by an experienced gastroenterologist. Finally, a novel risk stratification model or scoring system could not be proposed for neoplastic GB polyps because of data limitations. Instead, this study focused on identifying the clinical variables that could improve diagnostic accuracy. Future studies involving large-scale cohorts will be necessary to develop and validate a comprehensive risk stratification model.
In conclusion, the current surgical indications for GB polyps are still insufficient for the preoperative discrimination of neoplastic lesions. Nevertheless, combining CT and US, each with optimized criteria, can lead to more accurate treatment. Potential indications, such as sex, age, and platelet count, should be examined further to develop individualized risk models for neoplastic GB polyps.
SUPPLEMENTARY MATERIAL
Supplementary material is available at the Korean Journal of Gastroenterology website (https://www.kjg.or.kr/).
XML Download