图1 在IDEAL-IQ序列脂肪分量图像中进行PFVF测定
探讨磁共振胰腺的脂肪含量测定预测糖耐量异常(IGT)及2型糖尿病(T2DM)的可行性。
108名受试者(43男,65女;年龄47.9±12.1岁),包括T2DM(n=27),IGT(n=29)和葡萄糖耐量正常者(NGT;n=53)。NGT组进一步分为<40岁的NGT-年轻组和 ≥40岁的NGT-年长组。所有受试者均进行针对胰岛素抵抗和β细胞功能障碍的标准实验室检测。利用MRI非对称回波三点法水脂分离-定量(IDEAL-IQ)技术测定胰腺脂肪含量。分析胰腺脂肪体积分数(PFVF)与实验室检测参数之间的相关性,并分析PFVF预测IGT和T2DM的可行性。
T2DM患者的PFVF显著高于其余各组, IGT患者PFVF次之,NGT组PFVF最低。Logistic回归分析提示PFVF是血糖异常(IGT及T2DM)的独立危险因素。ROC曲线分析显示PFVF预测血糖异常(IGT及T2DM)的临界值为5.68%(P < 0.001, AUC:0.871, 敏感度92.7%,特异度71.7%)。
MRI脂肪定量测定技术可以测定胰腺的脂肪含量,为血糖异常(IGT和T2DM)的预测提供无创性的生物指标。
To investigate whether the quantitative magnetic resonance imaging (MRI) measurement of pancreatic fat content may indicate impaired glucose tolerance (IGT) or type 2 diabetes mellitus (T2DM).
A total of 108 subjects (43 men, 65 women; aged 47.9±12.1 years) were enrolled in this study, including 27 with T2DM, 29 with IGT and 53 with normal glucose tolerant (NGT). NGT individuals were subdivided into younger group (<40 years old) and elder group (≥40 years old). All the participants underwent standard laboratory tests for insulin resistance and β-cell dysfunction. MRI scanning with the iterative decomposition of water and fat with echo asymmetry and least square estimation image quantification (IDEAL-IQ) technique was used to determine fat distribution in the pancreas. The correlation between pancreatic fat volume fractions (PFVF) and laboratory parameters was analyzed. The feasibility of using PFVF to predict IGT and T2DM was evaluated.
PFVF was significantly higher in T2DM patients than that in both IGT and NGT patients, with lowest PFVF in NGT patients. Logistic regression indicated that PFVF was an independent risk factor of glucose abnormality (IGT and T2DM). ROC curve analysis showed that the cut-off value for PFVF to predict glucose abnormality (IGT and T2DM) was 5.68% (P< 0.001; AUC: 0.871; sensitivity: 92.7%; specificity: 71.7%).
MRI fat quantitative assay can quantitatively determine fat content in the pancreas, providing a non-invasive biomarker for the prediction of glucose abnormality (IGT and T2DM).
全球2型糖尿病(type 2 diabetes mellitus, T2DM)患者约3.87亿人,是威胁人类公共健康的主要疾病之一[
本研究共纳入2018年9月至2020年2月间108名受试者[(男43,女65;年龄(47.9±12.1)岁)],其中94例来自中山大学附属第一医院,9例来自江门市新会中医院,5例来自深圳市中医院。纳入标准:年龄18岁以上;无饮酒史或极少饮酒(以30度白酒为例,女性每天少于7.5 mL,男性每天少于11.25 mL酒精);血清铁蛋白<1 000 μg/ L;无肝脏或胰腺疾病(包括炎症、肿瘤、自身免疫性疾病等)。本研究通过上述医院医学伦理委员会批准,所有受检者检查前均签署知情同意书。
IGT和T2DM的诊断参考美国糖尿病协会2018年标准[
临床实验室指标检查项目:空腹血糖(fasting blood glucose,FBG),糖基化血红蛋白(glycosylated hemoglobin,HbA1c),体质量指数(body mass index,BMI),体脂含量(body fat content,BFC),甘油三酸酯(triglyceride,TG),低密度脂蛋白( low-density lipoprotein,LDL),高密度脂蛋白(high density lipoprotein,HDL), 总胆固醇(total cholesterol,CHOL),天冬氨酸转氨酶(aspartate aminotransferase,AST),丙氨酸转氨酶(alanine aminotransferase,ALT),空腹血浆胰岛素(fasting plasma insulin,FPI),稳态模型评估β细胞功能( homeostasis model assessment β,HOMA-β),稳态模型评估胰岛素抵抗(homeostasis model assessment of insulin resistance,HOMA-IR),定量胰岛素敏感性指数(quantitative insulin sensitivity index,QUICKI)和胰岛素作用指数(insulin action index,IAI)。
各中心均采用美国GE 公司生产的高场MR扫描仪,型号分别为:Brivo MR355,Signa Pioneer, Discorvery 750 w。扫描范围涵盖整个胰腺,扫描序列参数均保持一致,包括定位相及IDEAL-IQ序列。后者主要参数:TR 15.6 ms,一个TR采集6个回波,TE1 1.2~1.5 ms,其余回波时间以间隔2.0 ms递增,ETL 6,翻转角8°,层厚10 mm;仰卧位,检查前行呼吸训练,一次屏气扫描后即可得6组图像,分别为水相位、脂相图、脂肪分量图、弛豫率图、同相位图和反相位图。
采用GE MR副台工作站,由2名诊断医师及1名资深技师对图像进行分析,在IDEAL-IQ序列脂肪分量图像中进行胰腺脂肪含量测定,每位患者的脂肪含量取三位测量者测得数据的平均值。
胰腺脂肪含量测定:分别于胰头、体、尾选取3个圆形ROI,ROI面积为10 mm2,测量每个ROI脂肪分数,取均值作为全胰平均脂肪分数,ROI位置选定避开胰管和血管,并保证ROI 周围有胰腺实质环绕(
图1 在IDEAL-IQ序列脂肪分量图像中进行PFVF测定
Fig.1 Measurement of PFVF in fat fraction mapping of IDEAL-IQ sequence
Fat fraction mapping of the pancreas of a 46-year-old man from the IGT group. Circular ROIs (10 mm2) were manually drawn on the pancreas head (ROI-1), body (ROI-2) and tail (ROI-3).
采用IBM SPSS 20.0统计分析软件。评估每个受试者的胰头、体、尾之间的脂肪含量(fat volume fraction,FVF)差异,如果数据呈正态性分布、方差齐,采用方差分析;非正态分布、方差不齐采用非参数检验的Kruskal-Wallis H检验;使用Kruskal-Wallis H检验比较3组的胰腺FVF,差异有统计学意义时采用Bonferroni法进行两两比较。FVF与临床实验室指标测试结果之间的相关性分析采用Pearson相关分析法(正态分布数据)或Spearman相关分析法(计算等级数据之间的关系);采用logistics回归分析鉴定独立的危险因素,包括年龄,性别,FBG,HbA1c,BMI,BFC,PFVF,TG,LDL,CHOL,AST,ALT,FPI,HOMA-β ,HOMA-IR,QUICKI和IAI。以P <0.05为差异有统计学意义;利用特征性曲线(Relative operating characteristic curve,ROC curve)分析PFVF预测血糖异常(IGT及T2DM)的效能。
成功获取所有受试者的MR图像,采用Kendall’W检验分析3位观察者测量胰腺脂肪含量的一致性,结果显示3位观察者测量结果的Kendall’W系数为0.601 (P < 0.001),具有中等强度一致性。所有受试者的胰头[PFVF胰头:6.16 (4.83~8.65)]、体[PFVF胰体:6.45 (5.26~8.72)]、尾[PFVF胰尾:5.92 (4.17~8.00)]之间的PFVF差异均无统计学意义(
图2 胰腺中PFVF的箱形图
Fig.2 Box plots of PFVF in pancreas
PFVF values in the pancreas head, body, and tail. No significant difference was found among them (N=108,H=1.922,P = 0.383).The hollow points represent values that are 1.5 times more interquartile range than the upper quartile, while the star represent values that are 3 times more interquartile range than the upper quartile.
T2DM组、IGT组和NGT组之间比较:FBG(H=51.582,P <0.001;T2DM组与IGT组比较,P <0.001;T2DM组与NGT组比较,P <0.001),HbA1c(H=53.954,P <0.001;T2DM组与IGT组比较,P <0.001;T2DM组与NGT组比较,P <0.001),TG(H=33.4,P <0.001;T2DM组与IGT组比较,P <0.001;T2DM组与NGT组比较,P =0.001),LDL(F=11.651,P = 0.02;T2DM组与IGT组比较,P =0.343;T2DM组与NGT组比较,P =0.001),AST(H=8.658,P = 0.03;T2DM组与IGT组比较,P =0.756;T2DM组与NGT组比较,P <0.001),HOMA-β(H=26.158,P <0.001;T2DM组与IGT组比较,P <0.001;T2DM组与NGT组比较,P <0.001)和HOMA-IR(H=8.663,P = 0.03;T2DM组与IGT组比较,P =0.046;T2DM组与NGT组比较,P =0.021),以上指标T2DM组较其他两组高。HDL(H=8.647,P = 0.03;T2DM组与IGT组比较,P =0.086;T2DM组与NGT组比较,P =0.021)和QUICKI(H=41.497,P <0.001;T2DM组与IGT组比较,P <0.001;T2DM组与NGT组比较,P <0.001),以上指标T2DM组较其他两组低。3组的BMI、BFC、CHOL、ALT、FPI和IAI差异均无统计学意义(P > 0.05)。
PFVF与FBG、HbA1c、BMI、TG、CHOL、ALT、FPI和HOMA-IR呈正相关,与HDL、QUICKI和IAI呈负相关(
Clinical/lab results | PFVF | |
---|---|---|
rs | P | |
FBG | 0.379 | 0.005 |
HbA1c | 0.426 | <0.001 |
BMI | 0.358 | 0.003 |
TG | 0.535 | <0.001 |
LDL | 0.237 | 0.086 |
HDL | –0.292 | 0.002 |
CHOL | 0.419 | 0.003 |
AST | 0.187 | 0.119 |
ALT | 0.076 | 0.025 |
FPI | 0.298 | 0.02 |
HOMA β | 0.142 | 0.37 |
HOMA-IR | 0.309 | 0.002 |
QUICKI | –0.443 | 0.001 |
IAI | –0.323 | 0.003 |
rs: correlation coefficient. PFVF: fat volume fraction of pancreas; FBG: fasting blood glucose; HbA1c: glycosylated hemoglobin; BMI: body mass index; TG: triglyceride; LDL: low-density lipoprotein;HDL: high density lipoprotein; CHOL: total cholesterol; AST: aspartate aminotransferase; ALT: alanine aminotransferase; FPI: fasting plasma insulin; HOMA β: homeostasis model assessment β; IHOMA-IR: homeostasis model assessment of insulin resistance; QUICKI: quantitative insulin sensitivity index;IAI: insulin action index. PFVF correlated positively with BMI, HbA1c, FBG, CHOL, TG, ALT, FPI, and HOMA-IR.PFVF correlated negatively with HDL, IAI, and QUICKI.
T2DM受试者的胰腺脂肪含量最高,IGT受试者相对次之,NGT受试者为3组中最低(
图3 NGT、IGT和T2DM男性受试者的胰腺脂肪分量图
Fig.3 Represent fat fraction mappings of the pancreas of men in the NGT, IGT, and T2DM groups
A 20-year-old healthy NGT man with a mean PFVF of 2.7%. B: A 37-year-old man diagnosed as IGT, with a mean PFVF of 7.9%. C: A 59-year-old man who was first diagnosed as T2DM, with a mean PFVF of 16.1% .Fat content (correlated with signal intensity) is highest in the T2DM subject, less in the IGT subject, and least in the NGT subject.
在本研究中,NGT组被分为NGT年轻组和NGT年长组。NGT两个亚组、T2DM组及IGT组PFVF的组间比较应用Kruskal Wallis H检验,H=57.87,P < 0.001。两两比较采用Bonferroni法,NGT年长组的PFVF为5.58(4.61~7.58)明显高于NGT年轻组3.62(3.50~4.35),P < 0.001。IGT组的PFVF为7.15(5.82~8.81)显著高于NGT年长组(P = 0.002)。T2DM组的PFVF为8.81(6.61~12.77)明显高于IGT组(P = 0.014)和NGT年长组(P <0.001;
图4 T2DM、IGT、NGT-年长组和NGT-年轻组的箱形图
Fig.4 Box plots of T2DM, IGT, NGT-old and NGT-young groups
PFVF in NGT-old group were higher than that of the NGT-young group (P< 0.001), significant difference were found between the NGT group and IGT groups (P= 0.002). PFVF in the T2DM group were higher than that of the IGT (P= 0.014) and NGT groups (P< 0.001). The hollow points represent values that are 1.5 times more interquartile range than the upper quartile.
在Logistic回归中,我们将IGT及T2DM组合并,作为血糖异常组作为因变量,PFVF,FBG,HbA1c,TG,LDL,CHOL,FPI,HOMA-IR和QUICKI与因变量显著相关;年龄,性别,BMI,BFC,AST,ALT,HOMA-β及IAI与因变量无显著相关(
Variables | b | Sb | Wald χ2 | P | OR | OR 95% CI |
---|---|---|---|---|---|---|
PFVF | 0.823 | 0.162 | 25.809 | < 0.001 | 2.278 | (1.658, 3.129) |
FBG | 1.674 | 0.484 | 11.975 | 0.001 | 5.334 | (2.067,13.767) |
HbA1c | 2.891 | 0.717 | 16.231 | < 0.001 | 18.004 | (4.412,73.465) |
TG | 1.647 | 0.365 | 20.319 | < 0.001 | 5.193 | (2.537,10.628) |
LDL | 0.811 | 0.221 | 13.443 | < 0.001 | 2.250 | (1.459,3.471) |
CHOL | 0.665 | 0.214 | 9.636 | 0.002 | 1.944 | (1.278,2.958) |
FPI | 0.245 | 0.076 | 10.391 | 0.001 | 1.278 | (1.101,1.484) |
HOMA-IR | 0.568 | 0.245 | 5.212 | 0.0231 | 1.712 | (1.093,2.711) |
QUICKI | -9.021 | 1.990 | 20.554 | < 0.001 | < 0.001 | (<0.001,0.006) |
OR: odd ratio; PFVF: fat volume fraction of pancreas; FBG: fasting blood glucose; HbA1c: glycosylated hemoglobin; BMI: body mass index; BFC: body fat content; TG: triglyceride; LDL: low-density lipoprotein; HDL: high density lipoprotein; CHOL: total cholesterol; AST: aspartate aminotransferase; ALT: alanine aminotransferase; FPI: fasting plasma insulin; HOMA β: homeostasis model assessment β; HOMA-IR: homeostasis model assessment of insulin resistance; QUICKI: quantitative insulin sensitivity index; IAI: insulin action index. IGT and T2DM were significantly associated with PFVF, HbA1c, FBG, CHOL, TG, LDL, FPI, HOMA-IR, and QUICKI, but there was no significant association with age, gender, BMI, body fat content, ALT, AST, HOMA β, or IAI.
多元Logistic回归分析法分析了P <0.1的变量,包括年龄,FBG,HbA1c,TG,LDL,HDL,CHOL,ALT,FPI,HOMA-IR和QUICKI,未发现显著相关性。经过逐步回归分析后发现PFVF P = 0.008,OR = 1.615,OR 95%置信区间为(1.134,2.301)与IGT和T2DM的风险显著相关。利用特征性曲线(relative operating characteristic curve,ROC curve)分析PFVF预测IGT和T2DM的效能,其临界值为5.68%(P < 0.001, AUC: 0.871,敏感度92.7%,特异度71.7%;
图5 ROC曲线分析PFVF预测血糖异常(IGT及T2DM)的效能
Fig.5 ROC curve analysis PFVF predicts the effectiveness of IGT and T2DM
The critical value was 5.68% (P <0.001, AUC: 0.871, sensitivity: 92.7%, specificity: 71.7%).
由于MRI出色的软组织分辨率、无辐射及高精度等优势,已被广泛应用于非侵入性内部器官脂肪的检测或定量[
异位脂肪沉积是指多余的脂肪在肝脏、骨骼肌、心和胰腺等部位不良积聚,并与代谢性疾病密切相关。许多研究集中于酒精性脂肪肝中的肝脏脂肪含量,发现肝脏脂肪含量与BMI、LDL、TG和FBG密切相关。这些结果表明,肝脏中的脂肪积聚会影响脂质的代谢,并可能促进糖尿病的发展[
本研究中,我们不仅使用MRI IDEAL-IQ量化胰腺的脂肪含量,还分析了器官中的脂肪沉积模式。为了研究胰腺脂肪积聚与胰岛素抵抗及β细胞功能之间的相关性,我们评估了相对于NGT受试者的IGT和T2DM患者的胰腺脂肪含量。发现所有受试者的胰头、体和尾的脂肪含量无统计学差异,这表明胰腺中脂肪的堆积整体趋于均匀,与之前的报道一致[
本次研究中,我们探讨了胰腺的脂肪含量与临床实验室指标之间的相关性,这些参数证明了胰岛素抵抗和β细胞功能障碍。我们发现PFVF与BMI、TG、CHOL和ALT呈正相关。TG、CHOL和LDL是脂质代谢过程的重要参与者,它们在血液中的水平可以反映出肝脏的脂质处理功能[
Tushuizen等[
有研究[
根据目前的单因素Logistic回归分析,FBG、HbA1c、TG、LDL、CHOL、FPI、HOMA-IR和QUICKI是IGT和T2DM的独立危险因素。这些指标是临床评估和诊断糖耐量异常和胰岛素抵抗的参考。逐步回归分析则显示PFVF是IGT和T2DM的独立危险因素。Kruskal-Wallis检验还表明,NGT、IGT和T2DM组在FBG、HbA1c、TG、LDL、HDL、AST、HOMA-β、HOMA-IR和QUICKI方面存在显著差异。另一方面,年龄、性别、BFC、BMI、AST、ALT、HOMA-β和IAI则不是IGT和T2DM的统计学独立危险因素。
本研究的结果可能会受到样本量的限制,如BFC、BMI、AST和ALT等指标通常亦会受到其他疾病和状况的影响。此外,我们的结果表明,PFVF是IGT和T2DM的危险因素。但是,多变量分析中所有P值小于0.1的变量均显示阴性结果。可能的原因是,由于存在众多的风险因素均可影响IGT和T2DM的发展,并可能同时由多种因素共同构成了IGT和T2DM的风险。由于样本量限制,我们将IGT组与T2DM组合并为血糖异常组进行预测。PFVF是否能单独预测IGT及T2DM需要进一步扩大样本量研究。
总之,PFVF在IGT和T2DM的发展中具有重要作用,可作为IGT和T2DM的潜在风险指标。MRI可以无创定量PFVF,其可能是评估IGT和T2DM患者治疗后效果的理想指标, 在IGT和T2DM的预测及治疗方面均具有积极的临床意义。
Chng MH, Alonso MN, Barnes SE, et al. Adaptive immunity and antigen-specific activation in obesity-associated insulin resistance[J]. Mediators Inflamm, 2015, 20(15): 593075. [百度学术]
Bianchini JA, da Silva DF, Nardo CC, et al. Multidisciplinary therapy reduces risk factors for metabolic syndrome in obese adolescents[J]. Eur J Pediatr, 2013, 172(2): 215-221. [百度学术]
Heiskanen MA, Motiani KK, Mari A, et al. Exercise training decreases pancreatic fat content and improves beta cell function regardless of baseline glucose tolerance: a randomised controlled trial[J]. Diabetologia, 2018, 61(8): 1817-1828. [百度学术]
Nasr P, Fredrikson M, Ekstedt M, et al. The amount of liver fat predicts mortality and development of type 2 diabetes in non-alcoholic fatty liver disease[J]. Liver Int, 2020, 40(5): 1069-1078. [百度学术]
蔡华崧,翟凤仪,杨君,等.糖耐量正常试验者及糖耐量异常患者的全肝脂肪MRI定量研究[J].中山大学学报(医学科学版), 2015, 36(3): 396-402. [百度学术]
Cai HS, Zhai FY, Yang J, et al. MR quantification of total liver fat in patients with impaired glucose tolerance and with normal tolerance[J]. J Sun Yat-sen Univ (Med Sci), 2015, 36(3): 396-402. [百度学术]
王踉碕,彭屹峰. 3.0T MRI IDEAL-IQ序列评价成人肝铁浓度与脂肪分数的相关性[J].中国医学影像技术, 2015, 31(12): 1844-1847. [百度学术]
Wang LQ Peng YF. Correlation study of adult fat fraction and iron concentration with 3.0T MRI IdealIQ sequence[J]. Chin J Med Imag Technol, 2015, 31(12): 1844-1847. [百度学术]
American Diabetes Association. Improving Care and Promoting Health in Populations: Standards of Medical Care in Diabetes-2018[J]. Diabetes Care 2018, 41(Suppl 1): S7-12. [百度学术]
贾宏茹,李畅,刘畅,等. MR扩散峰度成像对非酒精性脂肪性肝病诊断价值[J]. 影像诊断与介入放射学, 2020, 29(1): 3-8. [百度学术]
Jia HR, Li C, Liu C, et al. Diagnostic value of non-alcoholic fatty liver disease with MR diffusion kurtosis imaging[J]. Diagn Imag & Intervent Radiol, 2020, 29(1): 3-8. [百度学术]
Patel NS, Peterson MR, Brenner DA, et al. Association between novel MRI-estimated pancreatic fat and liver histology-determined steatosis and fibrosis in non-alcoholic fatty liver disease[J]. Aliment Pharm Ther, 2013, 37(6): 630-639. [百度学术]
Livingstone RS, Begovatz P, Kahl S, et al. Initial clinical application of modified Dixon with flexible echo times: hepatic and pancreatic fat assessments in comparison with 1H MRS[J]. MAGMA, 2014, 27(5): 397-405. [百度学术]
Ramkissoon R, Gardner TB. Pancreatic steatosis: An emerging clinical entity[J]. Am J Gastroenterol, 2019, 11(4): 1726-1734. [百度学术]
Cohen JC, Horton JD, Hobbs HH. Human fatty liver disease: Old questions and new insights[J]. Science, 2011, 332(6037): 1519-1523. [百度学术]
Tushuizen ME, Bunck MC, Pouwels PJ, et al. Pancreatic fat content and beta-cell function in men with and without type 2 diabetes[J]. Diabetes Care, 2007, 30(11): 2916-2921. [百度学术]
Kahn SE, Hull RL, Utzschneider KM. Mechanisms linking obesity to insulin resistance and type 2 diabetes[J]. Nature, 2006, 444(7121): 840-846. [百度学术]
Hannukainen JC, Borra R, Linderborg K, et al. Liver and pancreatic fat content and metabolism in healthy monozygotic twins with discordant physical activity[J]. J Hepatol, 2011, 54(3): 545-552. [百度学术]
Wang CY, Ou HY, Chen MF, et al. Enigmatic ectopic fat: prevalence of nonalcoholic fatty pancreas disease and its associated factors in a Chinese population[J]. J Am Heart Assoc, 2014, 3(1): e000297. [百度学术]
63
浏览量
316
下载量
1
CSCD
相关文章
相关作者
相关机构