Article Text

Original research
Predictive value of anthropometric and biochemical indices in non-alcoholic fatty pancreas disease: a cross-sectional study
  1. Yang Xiao1,
  2. Han Wang1,
  3. Lina Han1,
  4. Zhibin Huang1,
  5. Guorong Lyu1,2,
  6. Shilin Li1
  1. 1Department of Ultrasonography, Second Affiliated Hospital of Fujian Medical University, Quanzhou, China
  2. 2Department of Medicine, Quanzhou Medical College, Quanzhou, China
  1. Correspondence to Dr Shilin Li; lslqz{at}fjmu.edu.cn

Abstract

Objectives Triglyceride (TG), triglyceride-glucose index (TyG), body mass index (BMI), TyG-BMI and triglyceride to high-density lipoprotein ratio (TG/HDL) have been reported to be reliable predictors of non-alcoholic fatty liver disease. However, there are few studies on potential predictors of non-alcoholic fatty pancreas disease (NAFPD). Our aim was to evaluate these and other parameters for predicting NAFPD.

Design Cross-sectional study design.

Setting Physical examination centre of a tertiary hospital in China.

Participants This study involved 1774 subjects who underwent physical examinations from January 2016 to September 2016.

Primary and secondary outcome measures From each subject, data were collected for 13 basic physical examination and blood biochemical parameters: age, weight, height, BMI, TyG, TyG-BMI, high-density lipoprotein (HDL), low-density lipoprotein, total cholesterol, TG, fasting plasma glucose, TG/HDL and uric acid. NAFPD was diagnosed by abdominal ultrasonography. A logistic regression model with a restricted cubic spline was used to evaluate the relationship between each parameter and NAFPD. The receiver operating characteristic (ROC) curve was used to calculate the area under the curve for each parameter.

Results HDL was negatively correlated with NAFPD, height was almost uncorrelated with NAFPD and the remaining 11 parameters were positively correlated with NAFPD. ROC curve showed that weight-related parameters (weight, BMI and TyG-BMI) and TG-related parameters (TyG, TG and TG/HDL) had high predictive values for the identification of NAFPD. The combinations of multiple parameters had a better prediction effect than a single parameter. All the predictive effects did not differ by sex.

Conclusions Weight-related and TG-related parameters are good predictors of NAFPD in all populations. BMI showed the greatest predictive potential. Multiparameter combinations appear to be a good way to predict NAFPD.

  • pancreatic disease
  • body mass index
  • health informatics

Data availability statement

Data are available upon reasonable request.

http://creativecommons.org/licenses/by-nc/4.0/

This is an open access article distributed in accordance with the Creative Commons Attribution Non Commercial (CC BY-NC 4.0) license, which permits others to distribute, remix, adapt, build upon this work non-commercially, and license their derivative works on different terms, provided the original work is properly cited, appropriate credit is given, any changes made indicated, and the use is non-commercial. See: http://creativecommons.org/licenses/by-nc/4.0/.

Statistics from Altmetric.com

Request Permissions

If you wish to reuse any or all of this article please use the link below which will take you to the Copyright Clearance Center’s RightsLink service. You will be able to get a quick price and instant permission to reuse the content in many different ways.

Strengths and limitations of this study

  • Risk factors of non-alcoholic fatty pancreas disease (NAPFD) were analysed on a large scale of participants to obtain the single predictive power and combined predictive power of each factor in predicting NAPFD.

  • The lack of important predictors such as waist circumference and hip circumference in this study may reduce the completeness of NAFPD prediction.

  • In this study, only ultrasound was used as a diagnostic method for NAFPD, which has a good effect but may still lead to missed diagnosis.

  • This study included only the Chinese population, and the findings may not apply to other ethnic groups.

Introduction

In recent decades, diet structure, working style and exercise status have changed, and the population obesity rate shows a continuously increasing trend.1 When the body fat content exceeds the storage capacity of the adipose tissue, it will be stored in non-fatty tissues, such as the liver and pancreas.2 Non-alcoholic fatty liver disease (NAFLD) is the most common liver metabolic disease worldwide, affecting approximately one-quarter of adults and children.3 The relationship between non-alcoholic fatty pancreas disease (NAFPD) and NAFLD is complex. Some studies have shown that the former is a predictor of the latter,4 whereas others have shown that the latter is a risk factor for the former.5 However, all studies agree that obesity is a key factor leading to ectopic fat deposition.

Increasing evidence suggests that NAFPD is associated with diabetes, pancreatitis and pancreatic cancer.6–9 Some studies have further shown the difference in adipose tissue in the pancreas between chronic pancreatitis and acute pancreatitis.10 11 The adipose tissue in chronic pancreatitis is replaced by more fibrous tissue, which is like a compartment, separating adipocytes from the pancreatic parenchyma, and reducing the damage to the pancreas. Acute pancreatitis lacks fibrous tissue, so more adipose tissue has a greater impact on the severity of acute pancreatitis. Moreover, studies show that the prevalence of NAFPD is high, from 30% to 33%.12 13 Therefore, early identification of NAFPD is essential to reducing the various related risks. Tissue biopsy is the gold standard for diagnosing NAFPD, as is NAFLD. However, pancreas biopsy is rarely performed in clinical practice because of its invasive nature, poor acceptability,14 and the procedural difficulty of the retroperitoneal position of the pancreas.

Some anthropometric and biochemical indicators, combinations of plain arithmetic and complicated numerical structures have been used to appraise the risk of NAFLD.15 16 Parameters previously used to assess NAFPD risk include triglyceride (TG) level, triglyceride-glucose index (TyG), body mass index (BMI), TyG-BMI and triglyceride to high-density lipoprotein ratio (TG/HDL).17–19 The present study was aimed to evaluate the best parameters to predict NAFPD through an epidemiological survey of 1774 general public subjects who underwent health checkups, including easy-collected anthropometric and biochemical indicators, combinations of plain arithmetic and complicated numerical structures.

Methods

Case data

1774 healthy subjects were selected from the Physical Examination Center of the Second Affiliated Hospital of Fujian Medical University. The baseline characteristics of the study participants are presented in table 1.

Table 1

Baseline characteristics of the study subjects with and without NAFPD

Data collection and calculations

Haematological samples were collected in the morning after at least 8 hours of fasting. Basic parameters such as age, height and weight and blood biochemical parameters such as high-density lipoprotein (HDL), low-density lipoprotein (LDL), total cholesterol (TC), TG, fasting plasma glucose (FPG), and uric acid (UA) were obtained. The composite parameters were calculated as follows: BMI=weight (kg)/height2 (m2), TyG=ln (fasting triglyceride (mg/dL) × fasting glucose (mg/dL)/2), TyG-BMI=TyG × BMI and TG/HDL=TG (mmol/L)/HDL (mmol/L).

Diagnosis and grading of NAFPD

NAFPD was diagnosed by ultrasound detection of pancreatic steatosis, ruling out medications, infections or alcohol as potential causes. Pancreatic ultrasonography was performed by two skilled technicians on the same morning blood was drawn. Patients were fasting until the completion of the ultrasound examination. At the time of the ultrasound examination, the patient was placed in the supine position with both arms raised above his head to fully expose the upper abdomen. The ultrasound machine and probe used were as follows: HI VISION Preirus (Hitachi (Japan)) with a probe C715 and SonoScape (China) with a probe C1-6A. Experienced senior physicians judged the ultrasound images without knowledge of the subjects’ medical information or biochemistry results. The diagnostic criterion for NAFPD was that the echo of the pancreas was enhanced and higher than that of the kidney cortex. Since the pancreas and kidney are often difficult to display on the same screen, the echo differences between the liver and kidney were compared first, followed by the comparison of the liver and pancreas.13 20 A schematic of this process is shown in figure 1.

Figure 1

Schematic diagrams of the pancreas under ultrasound. (A) Liver (blue circle), pancreas (red circle) and right renal cortex (orange circle) of a normal subject. The echo intensity of the pancreas was comparable to that of the right renal cortex. Liver echo was used as an intermediate bridge to better distinguish the difference between the pancreas and the right renal cortex. (B) Liver (blue circle), pancreas (red circle) and right renal cortex (orange circle) of an NAFPD subject. The echo intensity of the pancreas was significantly higher than that of the right renal cortex. NAFPD, non-alcoholic fatty pancreas disease.

On the basis that NAFPD had already been diagnosed, we used the following criteria to grade NAFPD.21–23 Grade 1 (mild): The echo intensity was slightly greater than that of the right renal cortex, and the pancreatic borders and splenic vein were clearly visible. Grade 2 (moderate): The echo intensity was significantly higher than that of the right renal cortex, but lower than that of the retroperitoneal fat, and the pancreatic borders were vague. Grade 3 (severe): The echo intensity was the same or higher than that of retroperitoneal fat, the pancreatic borders could not be evaluated and the splenic vein was not seen.

Statistical analysis

Data were analysed using SPSS V.26.0 software and R software (V.4.3.1). The distribution patterns of continuous variables were examined using QQ plots and Shapiro-Wilk tests. Continuous variables with a normal or approximately normal distribution were expressed as mean (SD) and compared using Student’s t-test, whereas continuous variables with skewed distribution were expressed as the median (quartile range) and compared using the Mann-Whitney U test. Categorical variables were expressed as numbers (%) and compared using Pearson’s χ2 test. A logistic regression model was established using restricted cubic splines (RCS), and the ORs and 95% CIs of NAFPD under different parameters were plotted. Four knots were placed at the 5th, 35th, 65th and 95th percentiles. In addition, to compare the predictive ability of the 13 parameters for NAFPD, receiver operating characteristic (ROC) curve analysis was used to calculate the largest sum of sensitivity and specificity and to determine the best threshold value for each parameter. The area under the curve (AUC) was handled according to the following criteria: <0.5, invalid; 0.5–0.65, poor; 0.65–0.75, moderate; 0.75–0.85, good; 0.85–1.0, very good. All tests were two-sided, and statistical significance was set at p<0.05.

Patient and public involvement

None.

Results

Characteristics of the subjects

A total of 928 female and 846 male subjects were included in this study. Among them, 417 women (44.94%) and 423 men (50.0%) were diagnosed with NAFPD. Table 1 describes the basic physical examination parameters, blood biochemical parameters and related composite parameters for the subjects diagnosed with and without NAFPD. Age (t=9.059 for women, t=9.127 for men, p<0.001), weight (t=16.539 for women, t=16.265 for men, p<0.001), BMI (t=21.953 for women, t=20.032 for men, p<0.001), TyG (t=12.506 for women, t=12.038 for men, p<0.001), TyG-BMI (t=18.732 for women, t=17.770 for men, p<0.001), LDL (t=8.392 for women, t=7.323 for men, p<0.001), TC (t=9.369 for women, t=8.172 for men, p<0.001), TG (u=11.991 for women, u=11.848 for men, p<0.001), FPG (u=7.824 for women, u=7.950 for men, p<0.001), TG/HDL (u=11.967 for women, u=11.794 for men, p<0.001) and UA (t=8.920 for women, t=6.169 for men, p<0.001) in patients with NAFPD were significantly higher than those in patients without NAFPD in both women and men. However, HDLs in patients with NAFPD were significantly lower than those in patients without NAFPD in both women and men (t=6.759 for women, t=7.417 for men, p<0.001). There was no significant difference in height between patients with NAFPD and without NAFPD in either women or men (t=0.472, p=0.631 for women; t=1.275, p=0.203 for men).

Associations between parameters and NAFPD

Figure 2 shows the associations between 12 parameters (except for height) and NAFPD risk in women (figure 2A1–L1) and men (figure 2A2–L2). Age, weight, BMI, TyG, TyG-BMI, LDL, TC, TG, FPG, TG/HDL and UA were positively correlated with NAFPD, whereas HDL was negatively correlated with NAFPD (all p<0.001). There was little difference between women and men. The difference lies only in the linear (p for non-linear >0.05) or non-linear (p for non-linear<0.05) correlations between each parameter and NAFPD. Data for height, which did not show any significant difference between patients with NAFPD and without NAFPD, are shown in online supplemental figure 1.

Figure 2

Dose-response relationship between NAFPD and 12 parameters (A) age, (B) weight, (C) BMI, (D) TyG, (E) TyG-BMI, (F) HDL, (G) LDL, (H) TC, (I) TG, (J) FPG, (K) TG/HDL and (L) UA. A1–L1 represent women, while A2–L2 represent men. BMI, body mass index; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; NAFPD, non-alcoholic fatty pancreas disease; TC, total cholesterol; TG, triglyceride; TG/HDL, triglyceride-to-high-density lipoprotein ratio; TyG, triglyceride-glucose index; TyG-BMI, triglyceride glucose-body mass index; UA, uric acid.

Correlation between parameters and grading of NAFPD

Table 2 describes the correlation between parameters and grading of NAFPD. Using grade 0 as the reference, the higher the grading of NAFPD, the higher the correlation (age, weight, BMI, TyG, TyG-BMI, TC, TG, FPG, TG/HDL and UA) (all p<0.001). Besides, using grade 0 as the reference, the higher the grading of NAFPD, the lower the correlation (HDL and LDL) (all p<0.001). Furthermore, there was no correlation between height and grading of NAFPD (all p>0.05).

Table 2

Correlation between each parameter and grading of NAFPD

Accuracy of each parameter in predicting NAFPD in the general public

ROC curve analysis was used to evaluate the accuracy of the 13 parameters for predicting NAFPD in the general public. The ability of the weight-related parameters (weight, BMI and TyG-BMI) to predict NAFPD was relatively good (table 3). The AUC for BMI was the largest (0.8526), with a sensitivity of 0.7774, specificity of 0.7752, best threshold of 22.3457, positive likelihood ratio of 3.4575 and negative likelihood ratio of 0.2872. The predictive power of the TG-related parameters (TyG, TG and TG/HDL) was slightly lower than that of the weight-related parameters; however, the AUC was greater than 0.70. The AUCs for age, HDL, LDL, TC, FPG and UA were all greater than 0.60. However, the AUC for height was only 0.5258.

Table 3

The best threshold, LR+, LR−, sensitivities, specificities and area under the curve of related indices for the screening of NAFPD in the general public

Accuracy of 13 parameters in predicting NAFPD by sex

Figure 3A,B and online supplemental table 1 show the results of the ROC curve analyses and AUCs for the 13 parameters used to predict NAFPD in women and men. The AUCs for all 13 parameters were greater than 0.5, indicating that all parameters had a predictive value for NAFPD. Additionally, the AUC for each parameter was similar between women and men, with no significant differences. Among all parameters, weight-related parameters had the largest AUCs. For women, AUCs were 0.8497 (95% CI: 0.8249 to 0.8746) for BMI, 0.8158 (95% CI: 0.7882 to 0.8434) for TyG-BMI and 0.7893 (95% CI: 0.7598 to 0.8188) for weight. The best thresholds for BMI, TyG-BMI and weight in women for predicting NAFPD were 23.2405, 52.9662 and 59.7500, respectively. For men, AUCs were 0.8580 (95% CI: 0.8330 to 0.8831) for BMI, 0.8186 (95% CI: 0.7907 to 0.8466) for TyG-BMI and 0.7946 (95% CI: 0.7649 to 0.8243) for weight. The best thresholds for BMI, TyG-BMI and weight in men for predicting NAFPD were 22.5014, 54.5370 and 62.2500, respectively. TG-related parameters also showed high NAFPD prediction performance, with AUCs greater than 0.70. The AUCs of the other parameters were greater than 0.60, except for that of height, which was close to 0.50.

Figure 3

ROC curve analysis of NAFPD-related indicators. (A) ROC curve analysis for women, (B) ROC curve analysis for men, (C) ROC curve analysis of multiple parameters. AUC, area under the curve; BMI, body mass index; FPG, fasting plasma glucose; HDL, high-density lipoprotein; LDL, low-density lipoprotein; M1, multiple parameters of age and weight for predicting NAFPD; M2, multiple parameters of age, weight and BMI; M3, multiple parameters of age, weight, BMI and TyG; M4, multiple parameters of age, weight, BMI, TyG and TyG-BMI; M5, multiple parameters of age, weight, BMI, TyG, TyG-BMI and HDL; M6, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL and LDL; M7, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL, LDL and TC; M8, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC and TG; M9, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC, TG and FPG; M10, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC, TG, FPG and TG/HDL; M11, multiple parameters of age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC, TG, FPG, TG/HDL and UA; NAFPD, non-alcoholic fatty pancreas disease; ROC, receiver operating characteristic; TC, total cholesterol; TG, triglyceride; TG/HDL, triglyceride-to-high-density lipoprotein ratio; TyG, triglyceride-glucose index; TyG-BMI, triglyceride glucose-body mass index; UA, uric acid.

Accuracy of multiple parameters in predicting NAFPD

Age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC, TG, FPG, TG/HDL and UA were numbered from 1 to 12. The combinations of multiple parameters were partly listed in online supplemental table 2. The AUC was larger when multiple parameters were combined (figure 3C). The AUC of the best model should be 0.8773, with a Youden index of 0.6109, positive likelihood ratio of 5.0469.

Discussion

In this study, 13 non-invasive parameters were evaluated for their associations with NAFPD and their ability to predict NAFPD. Age, weight, BMI, TyG, TyG-BMI, HDL, LDL, TC, TG, FPG, TG/HDL and UA had certain predictive effects on NAFPD, whereas the predictive effect of height on NAFPD was very limited. Notably, weight-related parameters had the best predictive performance, followed by TG-related parameters. Moreover, the predictive effects of these parameters did not differ between the sexes.

BMI is an indicator of obesity and is linked to an elevated risk of insulin resistance (IR), NAFLD or other metabolic diseases.24 25 The BMI cut-off in our study was <25 kg/m2, which is lower than the 30 kg/m2 cut-off in most Western countries, probably because of the higher intake of unsaturated fats in the Chinese population.26 The TyG, based on fasting blood glucose and triglyceride levels, is extensively used as an important indicator of cardiovascular events and IR.27–29 The above weight-related and TG-related parameters all had good predictive effects on NAFPD, with AUCs greater than 0.70. Although the accurate diagnosis of NAFPD requires a tissue biopsy, this is difficult to perform and runs counter to the Declaration of Helsinki. Therefore, the diagnosis often relies on indicators that are easy to collect, low-risk and highly accurate in practical clinical work. The weight-related and TG-related parameters in this study only required simple measurement and drawing of a small amount of peripheral blood to achieve a high NAFPD prediction effect. Therefore, the utility of these parameters should be promoted for surveillance in the general public to reduce the occurrence of major related diseases.

A further six parameters in this study (age, HDL, LDL, TC, FPG and UA) were also predictive of NAFPD, but to a lesser extent than the weight-related and TG-related parameters. Height has no obvious effect on the prediction of NAFPD, but the composite index formed after combining height with other indicators has a fairly broad predictive ability and excellent predictive performance,30–33 so it cannot be ignored. Additionally, the combinations of multiple parameters had higher prediction power and positive likelihood ratio than a single parameter. Multiparameter combination may be a good way to predict NAFPD.

In addition, we explored the correlation of each parameter with the grading of NAFPD. Some parameters, including age, weight, BMI, TyG, TyG-BMI, TC, TG, FPG, TG/HDL and UA, became more and more important as the grading of NAFPD increased. This means that in moderate-to-severe NAFPD, the above parameters are more worthy of detection. In contrast, the effect of HDL and LDL was more obvious only in the prediction of mild NAFPD, while it was not very satisfactory in the prediction of moderate-to-severe NAFPD.

Our study has several limitations. First, the study lacked information on some measurements, such as waist circumference (WC) and hip circumference. Thus, waist-to-height ratio (WHtR), TyG-WC, TyG-WHtR, conicity index, body roundness index, body-shape index, lipid accumulation product, visceral adiposity index, abdominal volume index and body adiposity index could not be calculated.34 35 The inclusion of these parameters may lead to a more accurate prediction of NAFPD. Second, the predictive effect of IR on NAFPD is missing in this study, which needs to be supplemented in the future. Third, the diagnosis of NAFPD was based solely on ultrasonographic findings. Many studies have proven the feasibility of ultrasound in the diagnosis of NAFPD.13 20 36 However, further research is required to confirm its accuracy. Furthermore, the data used in this study were from a Chinese population; therefore, the conclusions may not apply to other racial groups. Finally, the cross-sectional design employed in this study limits our interpretation of the causality between parameters.

Conclusions

In conclusion, our study showed that weight-related and TG-related parameters were good predictors of NAFPD, with BMI having the greatest predictive potential. Besides, multiparameter combinations may be a good way for better prediction for NAFPD.

Data availability statement

Data are available upon reasonable request.

Ethics statements

Patient consent for publication

Ethics approval

This study was conducted in full conformance with the principles of the Declaration of Helsinki, and the animal study protocol was approved by the Ethics Committee of the Second Affiliated Hospital of Fujian Medical University (protocol code 231 and date of approval 2022).

Acknowledgments

We would like to thank Editage (www.editage.cn) for English language editing.

References

Supplementary materials

  • Supplementary Data

    This web only file has been produced by the BMJ Publishing Group from an electronic file supplied by the author(s) and has not been edited for content.

Footnotes

  • Contributors The original draft of the manuscript was written by YX. The data were collected by HW, LH and ZH. The data curation and formal analysis were performed by YX and SL. The revision of the manuscript was done by YX, GL and SL. The funding acquisition was performed by SL. The responsibility for the accuracy of the data in this article and taking full responsibility for the work and/or conduct of the study, having access to the data, and controlling publication decisions are guaranteed by SL. All authors read and approved the final manuscript.

  • Funding This study was supported by the Quanzhou High-level Talents Innovation and Entrepreneurship Project under Grant number 2021C045R and the Joint funds for the innovation of science and technology, Fujian province under Grant number 2023Y9231.

  • Competing interests None declared.

  • Patient and public involvement Patients and/or the public were not involved in the design, or conduct, or reporting, or dissemination plans of this research.

  • Provenance and peer review Not commissioned; externally peer reviewed.

  • Supplemental material This content has been supplied by the author(s). It has not been vetted by BMJ Publishing Group Limited (BMJ) and may not have been peer-reviewed. Any opinions or recommendations discussed are solely those of the author(s) and are not endorsed by BMJ. BMJ disclaims all liability and responsibility arising from any reliance placed on the content. Where the content includes any translated material, BMJ does not warrant the accuracy and reliability of the translations (including but not limited to local regulations, clinical guidelines, terminology, drug names and drug dosages), and is not responsible for any error and/or omissions arising from translation and adaptation or otherwise.