Article Text
PDF
Abstract
Objective Type 2 diabetes is an important risk factor for non-alcoholic fatty liver disease (NAFLD), but current guidelines provide conflicting recommendations on whether diabetic patients should be screened for NAFLD. We therefore studied the strategy of screening diabetic patients by FibroScan.
Design Liver fat and fibrosis were assessed by controlled attenuation parameter (CAP) and liver stiffness measurements (LSM) by FibroScan at a diabetic centre for patients from primary care and hospital clinics. Probe-specific LSM cut-offs were used to detect advanced fibrosis.
Results Of 1918 patients examined, 1799 (93.8%) had valid CAP and 1884 (98.2%) had reliable LSM (1770 with the M probe and 114 with the XL probe). The proportion of patients with increased CAP and LSM was 72.8% (95% CI 70.7% to 74.8%) and 17.7% (95% CI 16.0% to 19.5%), respectively. By multivariable analysis, female gender, higher body mass index, triglycerides, fasting plasma glucose and alanine aminotransferase (ALT) and non-insulin use were associated with increased CAP. Longer duration of diabetes, higher body mass index, increased ALT and spot urine albumin:creatinine ratio and lower high-density lipoprotein-cholesterol were associated with increased LSM. Ninety-four patients (80% had increased LSM) underwent liver biopsy: 56% had steatohepatitis and 50% had F3-4 disease.
Conclusions Diabetic patients have a high prevalence of NAFLD and advanced fibrosis. Those with obesity and dyslipidaemia are at particularly high risk and may be the target for liver assessment. Our data support screening for NAFLD and/or advanced fibrosis in patients with type 2 diabetes.
Statistics from Altmetric.com
Video abstract
Significance of this study
What is already known on this subject?
Non-alcoholic fatty liver disease (NAFLD) is common in patients with type 2 diabetes.
Diabetes is an important risk factor of steatohepatitis and cirrhosis in NAFLD patients.
Current guidelines provide conflicting recommendations on NAFLD screening in diabetic patients because of the paucity of data.
It is possible to measure liver fat and fibrosis quickly by FibroScan. Its application as a screening tool in high-risk patients has not been systematically studied.
What are the new findings?
Around 70% of diabetic patients from primary care and hospital clinics had increased controlled attenuation parameter suggestive of NAFLD.
Around 18% of diabetic patients had increased liver stiffness.
Patients with high body mass index, dyslipidaemia and increased alanine aminotransferase were at highest risk of increased controlled attenuation parameter and liver stiffness.
Among patients with liver biopsy, 56% had steatohepatitis, 21% had advanced fibrosis and 29% had cirrhosis.
How might it impact on clinical practice in the foreseeable future?
Because of the high prevalence of NAFLD and advanced fibrosis, patients with type 2 diabetes may benefit from screening.
FibroScan is a reasonable initial assessment for patients with type 2 diabetes. However, its accuracy does not allow confident diagnosis of advanced disease. Further improvements in non-invasive assessment of liver fibrosis are needed.
Introduction
Non-alcoholic fatty liver disease (NAFLD) is the most common chronic liver disease, affecting 15%–40% of the population worldwide.1 Around 20%–30% of patients with NAFLD have non-alcoholic steatohepatitis (NASH), the active form of NAFLD which can cause liver fibrosis. This may eventually progress to cirrhosis and hepatocellular carcinoma in 10%–20% of patients.2–4
Type 2 diabetes is a major risk factor for NAFLD. Both feature insulin resistance as a core component of their pathophysiology. As such, up to 90% of diabetic patients in some populations also have NAFLD.5 However, due to insufficient data, current guidelines offer conflicting recommendations on whether diabetic patients should be screened for NAFLD.6 ,7 The main arguments against screening include uncertainties surrounding diagnostic tests and treatment options and lack of knowledge related to the long-term benefits of screening. However, screening may identify patients with NAFLD-related cirrhosis who would benefit from hepatocellular carcinoma and varices surveillance.
Knowledge on the epidemiology of NAFLD is incomplete because of the limitations of various diagnostic modalities. Liver biopsy is considered the reference standard, but is impractical to apply to a large study population. Abdominal ultrasonography is easily accessible but is only qualitative, poor in detecting minor steatosis and suffers from intraobserver and interobserver variability. Furthermore, it cannot assess disease severity.
Transient elastrography is a non-invasive test of liver fibrosis that is quick and easy to perform and has a high degree of patient acceptance.8 It has high accuracy and reproducibility when used to detect advanced fibrosis and cirrhosis. In addition, the latest model measures a novel physical parameter called the controlled attenuation parameter (CAP). Since fat affects ultrasound propagation, CAP measurement has been shown to be accurate in estimating the amount of liver fat.9–11 Using this non-invasive technique, it is now possible to measure liver fat and fibrosis in a large number of patients.
In this study, we aim to test the strategy of NAFLD and fibrosis screening in patients with type 2 diabetes. We also studied factors associated with increased CAP and liver stiffness to guide selection of patients for screening.
Methods
Subjects
This is a prospective cohort study. From March 2013 to May 2014, we screened 2466 consecutive patients aged ≥18 years with type 2 diabetes who attended comprehensive diabetic complications screening at the Diabetes Mellitus and Endocrine Centre, Prince of Wales Hospital, Hong Kong. The source of referrals included hospital and primary care clinics in the New Territories East Cluster, which serves a population of 1.2 million. Subjects with active malignancy, positive hepatitis B surface antigen or antibody against hepatitis C virus, secondary causes of fatty liver (eg, consumption of amiodarone and tamoxifen) and congestive hepatopathy were excluded. Men who consumed >20 g and women who consumed >10 g of alcohol per day were also excluded. All patients provided informed written consent. The patients would be prospectively followed up for 10 years for hepatic, cardiovascular and metabolic complications. This paper reports the results of the baseline hepatic assessment.
Clinical assessment
The patients underwent a comprehensive assessment for diabetes-related complications and risk factors according to the European DIABCARE protocol.12 The assessment included an interview by diabetes nurses, vitals and anthropometric measurements, fundus examination and podiatry assessment. The medical history, current drugs, smoking and alcohol consumption were recorded using standard questionnaires. Body mass index (BMI) was calculated as body weight (kg) divided by body height (m2). Waist circumference was measured at a level midway between the lower rib margin and iliac crest with the tape all around the body. Blood pressure was measured on both arms in the sitting position after resting for at least 15 min. After fasting for 8 h overnight, blood was sampled for assays of fasting lipids, glucose, glycated haemoglobin (HbA1c), renal and liver function tests. The upper limit of normal for alanine aminotransferase (ALT) was 30 IU/L for men and 19 IU/L for women.13 The abbreviated modification of diet in renal disease equation recalibrated for Chinese was used to calculate the estimated glomerular filtration rate (eGFR).14 Spot urine for albumin:creatinine ratio was performed to detect albuminuria.
FibroScan examination
During the diabetic complication assessment visit, liver stiffness measurement (LSM) and CAP were obtained using FibroScan502 (Echosens, Paris, France) as described previously.15 All patients were fasted for at least 8 h before the procedure. The LSM score was represented by the median of 10 measurements and was considered reliable only if at least 10 successful acquisitions were obtained and the IQR-to-median ratio of the 10 acquisitions was ≤0.3. The CAP score was represented by the median value. Because the meaning of IQR-to-median ratio for CAP is less well defined, CAP measurements were considered reliable and included in the final analysis if 10 successful acquisitions were obtained. The M probe was used in the first instance for all patients so that both LSM and CAP could be obtained. If the M probe failed, the XL probe catering for obese patients was used.16
Hepatic steatosis was graded by CAP using the M probe according to published cut-offs (S1=222–232; S2=233–289; S3 ≥290 dB/m).10 Probe-specific LSM cut-offs used to define advanced fibrosis and cirrhosis (M probe F3=9.6–11.4, F4 ≥11.5; XL probe F3=9.3–10.9, F4 ≥11.0 kPa) were derived from previous studies.15 ,16 Two operators performed the procedures. Both had >5 years of experience with FibroScan and had performed >2000 procedures.
Liver histology
Patients with suspected advanced fibrosis or cirrhosis based on the FibroScan examination were invited to undergo liver biopsy. Percutaneous liver biopsy was performed using the 16G Temno needle. Liver histology was assessed by a single experienced pathologist (AW-HC) who was blinded to the clinical data. Histological scoring was performed according to the NASH Clinical Research Network system.17 Fibrosis was staged from 0 to 4: F0=absence of fibrosis, F1=perisinusoidal or portal, F2=perisinusoidal and portal/periportal, F3=septal or bridging fibrosis and F4=cirrhosis. NASH was defined by the presence of hepatic steatosis and inflammation with hepatocyte injury (ballooning) with or without fibrosis.7
Statistical analysis
Data were summarised and presented using appropriate descriptive statistics. The normality of continuous variables was assessed by skewness statistic and graphically by normal probability plot. Triglycerides, fasting plasma glucose, HbA1c, plasma creatinine and ALT were all natural log-transformed before being entered into inferential statistical analysis. Patient characteristics between those with and without valid CAP measurements were compared using independent t test, χ2 or Fisher's exact tests as appropriate. Multivariable logistical regression analyses were conducted to identify patient characteristics independently associated with each of the outcomes: (1) NAFLD and (2) advanced fibrosis. Univariate analysis was first performed on each of the considered independent variables to select candidate variables for the multivariable analyses. Those factors with a p value <0.25 in the univariate analyses were selected as candidate variables for backward stepwise multivariable logistical regressions to delineate factors independently associated with each of the outcomes.18 Furthermore, to avoid collinearity, only the independent variable with the highest Wald statistic value (or equivalently the smallest p value) was selected as candidate variable among each set of highly intercorrelated (r>0.8) independent variables (including (1) body weight, BMI and waist circumference, (2) systolic and diastolic blood pressure, (3) total cholesterol and low-density lipoprotein (LDL) cholesterol). Hosmer–Lemeshow test was used to assess the goodness of fit of the final multivariable logistical regression models obtained.18 An insignificant result of Hosmer–Lemeshow test indicates a model that fits the data well. The results of significant factors identified were presented with their OR and 95% CIs. All the statistical analyses were performed using SPSS V.22.0 (SPSS Inc, Chicago, Illinois, USA). All statistical tests involved were two-tailed and statistical significant level was set at 0.05.
For a prevalence of NAFLD ranging from 30% to 90%, a sample size of 1800 subjects would estimate the prevalence of NAFLD with a margin of error ranging from 1.4% to 2.3% at 5% level of significance. For a prevalence of advanced fibrosis of 1%–20%, the same sample size would estimate the prevalence with margin of error ranging from 0.5% to 1.9% at 5% level of significance.
Results
A total of 2119 patients fulfilled the inclusion criteria and underwent FibroScan assessment, and 1918 underwent complete diabetes complication screening. Of these, 1884 (98.2%) had a reliable LSM by either the M probe or the XL probe, while 1799 (93.8%) had reliable CAP scores using the M probe (figure 1).
Study participant flow. CAP, controlled attenuation parameter; LSM, liver stiffness measurement.
The mean age was 61 years, and 54% were men (table 1). A total of 6.6% had platelet count <150×109/L (6.6% of the patients with valid CAP measurements and 6.7% of those without). Compared with patients who failed M probe examination, those with valid CAP data had shorter duration of diabetes, lower BMI, waist circumference, systolic blood pressure and HbA1c, were more likely to smoke and were less likely to be men and use antihypertensives and insulin. Patients with valid CAP also had less albuminuria.
Characteristics of the study population
Proportion of patients with increased CAP and LSM
Among 1799 patients with valid CAP data, the median CAP score was 266 dB/m (IQR 216–313). One thousand three hundred and nine (72.8%, 95% CI 70.7% to 74.8%) had increased CAP of ≥222 dB/m (figure 2A). The number of patients with grade 1, 2 and 3 steatosis as suggested by CAP values was 92 (5.1%), 533 (29.6%) and 684 (38.0%), respectively.
Distribution of (A) controlled attenuation parameter (n=1799), (B) liver stiffness by the M probe (n=1770) and (C) liver stiffness by the XL probe (n=114) in the entire cohort. The red lines represent the threshold for abnormal values: (A) 222 dB/m for controlled attenuation parameter, (B) 9.6 kPa for liver stiffness by the M probe and (C) 9.3 kPa for liver stiffness by the XL probe.
Among 1770 patients with reliable LSM by the M probe, the median LSM was 6.3 kPa (IQR 4.9–8.3). Three hundred and three of them (17.1%) had LSM ≥9.6 kPa by M probe suggestive of advanced fibrosis or cirrhosis; 199 (11.2%) also had LSM ≥11.5 kPa suggestive of cirrhosis (figure 2B). In addition, 114 patients failed M probe examination but had successful LSM by the XL probe. Their median LSM was 6.9 kPa (IQR 4.8–9.6). Thirty one of them (27.2%) had LSM ≥9.3 kPa by XL probe suggestive of advanced fibrosis or cirrhosis; 25 (21.9%) had LSM ≥11.0 kPa suggestive of cirrhosis (figure 2C). Taken together, the proportion of patients with increased LSM by either the M probe or the XL probe was 17.7% (95% CI 16.0% to 19.5%).
As expected, increased LSM was mainly observed in patients with increased CAP. Two hundred and sixty nine of 1309 (20.6%) patients with increased CAP had increased LSM, compared with 34 of 490 (6.9%) patients with normal CAP (p<0.001). The median LSM of patients with normal CAP but increased LSM was 14.2 kPa (IQR 11.1–16.5). Some of these patients might have burnt out NASH.
Factors associated with increased CAP
By univariate analysis, increased CAP ≥222 dB/m was associated with higher body weight, BMI, waist circumference, diastolic blood pressure, triglycerides, fasting plasma glucose and ALT (table 2). It was also associated with lower high-density lipoprotein (HDL)-cholesterol and the use of oral antidiabetic drugs, antihypertensives and lipid-lowering drugs. Fewer patients with increased CAP were on insulin treatment. By multivariable analysis, female gender, higher BMI, triglycerides, fasting plasma glucose and ALT level, and not using insulin remained as independent factors associated with increased CAP (Hosmer–Lemeshow test; p=0.564). Increased CAP was found in 410 of 751 (54.6%), 609 of 736 (82.7%) and 280 of 296 (94.6%) patients with BMI <25, 25–30 and ≥30 kg/m2, respectively (p<0.001; figure 3). When the different classes of oral antidiabetic drugs were analysed separately, none of them were associated with an increased or decreased risk of increased CAP in the multivariable analysis (data not shown).
Factors associated with increased controlled attenuation parameter
Proportion of patients with increased controlled attenuation parameter (CAP) and liver stiffness measurement (LSM) according to body mass index. A CAP value of ≥222 dB/m was used to estimate the proportion of patients with non-alcoholic fatty liver disease. An LSM of ≥9.6 kPa by the M probe or ≥9.3 kPa by the XL probe was used to estimate the proportion of patients with advanced fibrosis.
Two hundred and seventy six of 335 (82.4%) men with ALT ≥30 IU/L had increased CAP, compared with 434 of 665 (65.3%) men with ALT <30 IU/L (p<0.001). Similarly, 409 of 502 (81.5%) women with ALT ≥19 IU/L had increased CAP, compared with 185 of 291 (63.6%) women with ALT <19 IU/L (p<0.001). The sensitivity, specificity, positive predictive value and negative predictive value of the combined ALT cut-offs to detect NAFLD by CAP criteria were 52.5% (95% CI 49.8% to 55.3%), 68.9% (95% CI 64.6% to 73.0%), 81.8% (95% CI 79.1% to 84.4%) and 35.3% (95% CI 32.2% to 38.4%), respectively.
Factors associated with increased LSM
By univariate analysis, increased LSM (≥9.6 kPa by M probe or ≥9.3 kPa by XL probe) was associated with higher body weight, BMI, waist circumference, blood pressure, triglycerides, fasting plasma glucose, HbA1c, ALT and spot urine albumin:creatinine ratio (table 3). It was also associated with lower total cholesterol, HDL-cholesterol, LDL-cholesterol and eGFR and the use of antihypertensives and lipid-lowering drugs. By multivariable analysis, longer duration of diabetes, higher BMI, ALT and spot urine albumin:creatinine ratio and lower HDL-cholesterol remained as independent factors associated with increased LSM (Hosmer–Lemeshow test, p=0.672). Increased LSM was found in 60 of 744 (8.1%), 141 of 754 (18.7%) and 131 of 370 (35.4%) patients with BMI <25, 25–30 and ≥30 kg/m2, respectively (p<0.001; figure 3). When the different classes of oral antidiabetic drugs were analysed separately, none of them were associated with an increased or decreased risk of increased LSM in the multivariable analysis (data not shown).
Factors associated with increased liver stiffness
One hundred and eight of 345 (31.3%) men with ALT ≥30 IU/L had increased LSM, compared with 85 of 673 (12.6%) men with ALT <30 IU/L (p<0.001). Similarly, 116 of 540 (21.5%) women with ALT ≥19 IU/L had increased LSM, compared with 24 of 320 (7.5%) women with ALT <19 IU/L (p<0.001). The sensitivity, specificity, positive predictive value and negative predictive value of the combined ALT cut-offs to detect advanced fibrosis by LSM criteria were 67.3% (95% CI 61.9% to 72.3%), 57.2% (95% CI 54.7% to 59.7%), 25.3% (95% CI 22.5% to 28.3%) and 89.0% (95% CI 86.9% to 90.9%), respectively. The proportion of patients with increased CAP and LSM by ALT level is shown in table 4.
Proportion of patients with increased CAP and LSM by ALT levels
Liver histology
Ninety-four patients underwent liver biopsy (table 5). The mean liver biopsy length was 20 (SD 7) mm and mean number of portal tracts was 8 (SD 3). Eighty-seven patients had M probe examination with a median CAP of 320 dB/m (IQR 286–350 dB/m) and LSM 14.1 kPa (IQR 11.8–20.6 kPa). Seven patients only had XL probe examination, and the median LSM was 17.6 kPa (IQR 8.6–29.2 kPa). Seventy-five of these 94 (80%) patients had increased LSM (≥9.6 kPa by M probe or ≥9.3 kPa by XL probe). The sensitivity, specificity, positive predictive value and negative predictive value of increased LSM to detect histological F3-4 disease were 94%, 34%, 59% and 84%, respectively. Lobular inflammation was found in 89 (95%) patients, hepatocyte ballooning in 54 (57%) and fibrosis in 89 (95%; table 5). In particular, 60 (50%) patients had F3-4 disease. Seventy-eight (83%) patients had NAFLD activity score greater than or equal to three. Fifty-three (56%) subjects had NASH.
Histological severity of 94 patients with liver biopsy
Seven subjects had steatosis found in <5% of hepatocytes (liver specimen length 14–26 mm). Three had normal liver histology, one had mild lobular inflammation but no fibrosis, two had bridging fibrosis (F3) and one had cirrhosis (F4). All three patients with F3-4 disease had grade 1–2 lobular inflammation and no other aetiologies of chronic liver disease. The picture was suggestive of burnt out NASH.
Discussion
In this large prospective hospital cohort, the majority of diabetic patients had increased CAP suggestive of NAFLD. A significant proportion also had increased LSM. Patients with high BMI and additional metabolic factors were most likely to have increased CAP and LSM.
In previous community-based studies using abdominal ultrasonography, around 32%–62% of diabetic patients were found to have NAFLD.19–22 The prevalence of NAFLD is higher in our cohort for several reasons. First, abdominal ultrasonography is operator dependent and is insensitive to mild steatosis.23 Therefore, NAFLD may be underreported and missed in some cases. In comparison, CAP measurement by FibroScan can detect liver fat involving as little as 10% of the hepatocytes.9–11 Second, our study was conducted at the hospital setting. Although we had an open referral system for both hospital and primary care clinics, the metabolic burden of our patients was higher than that of patients in the community setting. Our findings therefore more reflect the situation at hospital clinics.
Since we could measure CAP and liver stiffness simultaneously with FibroScan, it was possible to assess the prevalence of NAFLD and also disease severity. According to past histological studies, diabetes is one of the most important risk factors of cirrhosis in patients with NAFLD.24 ,25 At the population level, diabetes also doubles the risk of hepatocellular carcinoma in men.26 A retrospective study of 1131 diabetes patients using FibroTest also reported a prevalence of between 2.8% and 5.6% for advanced fibrosis.27 Similarly, studies from Europe and Australia found that 5%–35% of diabetic patients had increased LSM by different cutoffs.28–30 In our study, 18% of the patients were found to have increased LSM. We further performed liver biopsy in 94 patients and confirmed the presence of advanced fibrosis or cirrhosis in half of the patients. If we adjust for the positive predictive value of LSM, around 10% of our diabetic patients would still have advanced liver disease. This argues strongly in favour of liver assessment in diabetic patients.
Previous studies report an LSM failure rate of around 3% and unreliable measurements in 12%–16%.31 ,32 In the current study, however, 98.2% of the subjects could have successful and reliable LSM. Differences in study design can explain the apparent discrepancies. First, the use of XL probe allowed successful LSM even in obese patients.16 Second, recent data suggest that the success rate of LSM (number of valid acquisitions divided by the total number of acquisitions) does not affect the reliability of LSM as compared with liver histology.33 Therefore, we did not include success rate as one of the reliability criteria in this study. Finally, all examinations were performed by experienced operators who had performed >2000 measurements before. Operator experience is pivotal in achieving successful and reliable LSM.31
Although FibroScan is easy to perform, it is unlikely that clinicians can apply it to all diabetic patients because of the large number of patients and the availability of assessment. Therefore, it is important to identify patients at risk of NAFLD and advanced liver disease. Traditional metabolic risk factors such as high BMI and dyslipidaemia were independent factors associated with NAFLD and increased LSM. In other words, patients with diabetes and additional metabolic diseases are at higher risk of having advanced liver disease and may benefit from liver assessment. That said, it is important to note that among diabetic patients with BMI <25 kg/m2, 55% still had increased CAP and 8% had increased LSM. In previous population studies, NAFLD and advanced fibrosis are observed in a small but significant proportion of non-obese subjects. Such patients usually have other components of metabolic syndrome and despite relatively normal BMI often have recent weight gain.1 ,34 In contrast, ALT correlates poorly with histological severity.35 ,36 Metabolic factors are more important than ALT in guiding liver assessment in diabetic patients.
Thiazolidinediones improve liver histology in NASH patients, and metformin use has been shown to be associated with a lower mortality rate in cirrhotic patients.37 ,38 Nevertheless, neither class of drugs was independently associated with NAFLD or advanced fibrosis in our study. It should however be noted that the current study is not a randomised controlled trial. The selection of different drugs was influenced by the underlying disease status. Unexpectedly, insulin use was associated with a lower risk of NAFLD. This is counterintuitive as patients requiring insulin treatment usually have failed treatment with oral antidiabetic drugs and thus have poorer diabetic control. One possible explanation is that patients with cirrhosis are more insulin resistant and therefore more likely require insulin treatment. As a NAFLD patient progresses to cirrhosis, liver fat tends to disappear and the patient would thus be classified as having no fatty liver.39 In fact, there was a trend that insulin use was associated with advanced fibrosis in our entire cohort (table 3). Three of our patients who underwent liver biopsy also had features of burnt out NASH.
Our study has the strength of a large sample size and the use of one of the best and widely available non-invasive tests of liver steatosis and fibrosis. Compared with previous studies, ours also had comprehensive diabetic assessment. Nevertheless, there were several limitations. First, at the time of the study, CAP could only be measured by the M probe. More obese patients who required XL probe examination only had LSM but not CAP data. The true prevalence of NAFLD would therefore be even higher. This would be particularly relevant to Western countries where the prevalence of obesity is higher. In the Caucasian cohorts described above, 8%–15% of the diabetic patients failed M probe examination.28–30 Nonetheless, the FibroScan programme has been updated to allow CAP measurement by XL probe. If validated, CAP measurement can also be done in obese patients. Third, liver biopsy was only performed in a subset of patients. However, it is unethical to biopsy patients with no apparent liver disease. A low LSM also has excellent negative predictive value in excluding advanced fibrosis.15 Along the same line, the diagnostic performance of LSM in the histology subgroup of this study should be interpreted with caution because patients undergoing liver biopsy were selected based on increased likelihood of advanced disease. Finally, we only reported the baseline assessment. The patients are currently under prospective follow-up for 10 years. The prognostic implications of the baseline liver assessment will be further unravelled.
In conclusion, diabetic patients at hospital and primary care clinics have a high prevalence of NAFLD and advanced liver fibrosis. Diabetic patients with high BMI and dyslipidaemia are at particularly high risk and may be the target for liver assessment. Our data support screening for NAFLD and/or advanced fibrosis in patients with type 2 diabetes.
References
Footnotes
Contributors Study inception: VW-SW; data collection: RK, GL-HW, YZ, AO-YL, SS-TS, AW-HC, M-WY and VW-SW; data analysis: KCC and VW-SW; administrative support: HL-YC, JC-NC and AP-SK; manuscript drafting: RK and VW-SW; All authors read and approved the final version of the manuscript.
Funding This study was supported by a grant from the Research Grants Council, the Hong Kong SAR Government (Project reference CUHK477813).
Competing interests GL-HW, HL-YC and VW-SW have served as speakers for Echosens. JC-NC has received research grant and/or honorarium for consultancy or giving lectures from Bayer, Boehringer Ingelheim, Daiichi-Sankyo, Eli-Lilly, GlaxoSmithKline, Merck Sharp and Dohme, Merck Serono, Pfizer, Astra Zeneca, Sanofi, Novo-Nordisk and/or Bristol-Myers Squibb. AP-SK has received honorarium for consultancy or giving lectures from Abbott, Astra Zeneca, Sanofi, Novo-Nordisk, Eli-Lilly, Merck Serono, Pfizer and Nestle. Part of the proceeds has been donated to The Chinese University of Hong Kong, American Diabetes Association and other charity organisations to support diabetes research and education. Other authors declared no competing interests with this manuscript.
Ethics approval Clinical Research Ethics Committee, The Chinese University of Hong Kong.
Provenance and peer review Not commissioned; externally peer reviewed.