Abstract
Background:
We examined the association of objective measures of cardiometabolic risk with progression to a high-risk for advanced fibrosis in patients with MASLD at initially low- and indeterminate-risk for advanced fibrosis.
Methods:
We performed a retrospective cohort study of primary care patients with MASLD between 2012 and 2021. We evaluated patients with MASLD and low- or indeterminate-risk Fibrosis-4 Index (FIB-4) scores and followed them until the outcome of a high-risk FIB-4 (≥2.67), or the end of the study period. Exposures of interest were body mass index (BMI), systolic blood pressure (SBP), hemoglobin A1c, cholesterol, estimated glomerular filtration rate (eGFR), and smoking status. Variables were categorized by the threshold for primary care therapy intensification. Unadjusted and adjusted Cox regression models were developed for the outcome of time to a high-risk FIB-4 value.
Results:
The cohort included 1,347 patients with a mean follow-up of 3.6 years (SD 2.7). Of the cohort, 258 (19%) had a subsequent FIB-4 ≥ 2.67. In the fully adjusted Cox regression models, mean SBP ≥ 150 mm Hg (1.57; 95%CI 1.02–2.41) and eGFR ≤ 59 ml/min (HR 2.78; 95%CI 2.17–3.58) were associated with an increased hazard of a high-risk FIB-4, while receiving a statin prescription (HR 0.51; 95%CI 0.39–0.66) was associated with a lower risk.
Conclusions:
Nearly 1 in 5 primary care patients with MASLD transitioned to a high-risk FIB-4 score during 3.6 years of follow-up, and uncontrolled blood pressure and reduced kidney function were associated with an increased hazard of a FIB-4 at high-risk for advanced fibrosis.
Keywords: FIB-4, Fibrosis-4 Index, metabolic dysfunction-associated steatotic liver disease, nonalcoholic fatty liver disease, NAFLD
Introduction
Metabolic dysfunction-associated steatotic liver disease (MASLD, formerly nonalcoholic fatty liver disease [NAFLD]) affects an estimated 30% of the U.S. population, is a major contributor to the growing burden of chronic liver disease, and is a leading cause of cirrhosis, hepatocellular carcinoma, and need for liver transplantation (1–4). As a significant, and growing, public health concern, management of MASLD requires the collaboration of hepatology, endocrinology, and primary care clinicians to improve detection and public awareness, initiate therapeutic weight loss efforts, and identify complications of the disease (3, 5–7).
Recent MASLD management guidelines highlight the significant role of primary care in combatting the MASLD epidemic. Specifically, recommendations focus on detection of advanced fibrosis by primary care clinicians (3, 5). Advanced fibrosis is the single best predictor of future severe liver outcomes, including cirrhosis and hepatocellular carcinoma, in patients with MASLD (8, 9). In the guidelines, primary care clinicians can assess advanced fibrosis using a combination of the Fibrosis-4 Index (FIB-4) and a confirmatory test, consisting of either a liver stiffness measurement (LSM) with vibration-controlled elastography (VCTE) or the Enhanced Liver Fibrosis (ELF) test, a proprietary blood test approved as a prognostic biomarker for advanced fibrosis (10, 11). Patients with a FIB-4 at indeterminate (FIB-4 ≥ 1.3 in patients younger than 65 years or FIB-4 ≥ 2.0 in patients age 65 and older [this higher cut-off has demonstrated higher specificity and a lower false positive rate in older patients]) or high-risk for advanced fibrosis (FIB-4 ≥ 2.67) require a confirmatory assessment with VCTE or ELF (11, 12). The guidelines recommend that those patients with a confirmatory VCTE (LSM ≥ 8 kPa) or ELF (ELF score ≥ 9.8) at high-risk for advanced fibrosis should receive a referral to a hepatology specialist and those with either a low-risk FIB-4 (FIB-4 < 1.3) or a low-risk confirmatory test (VCTE LSM < 8 kPa; ELF < 9.8) should remain in the primary care setting (3, 5, 11).
How primary care clinicians should manage MASLD at low-risk for advanced fibrosis requires further clarity. In the absence of a specific, approved pharmacotherapy for MASLD, weight loss and cardiovascular risk reduction have been the keystones for reducing steatosis and preventing atherosclerotic cardiovascular events, respectively (1, 11, 13). With advanced fibrosis playing a pivotal role in the prognosis of MASLD, providing primary care targets for preventing progression of advanced fibrosis risk would significantly contribute to primary care management (14). In this study, we examined the association of objective measures of cardiometabolic risk factors, commonly addressed in primary care, with progression to a high-risk for advanced fibrosis in patients with MASLD and an initial low- or indeterminate-risk for advanced fibrosis. We hypothesized that one or more of the cardiometabolic measures studied would be associated with an increased hazard of fibrosis risk progression which could provide a future therapeutic target for primary care MASLD management.
Materials and Methods
We conducted a retrospective cohort study of primary care patients with MASLD at initially low- or indeterminate-risk for advanced fibrosis. Patients with a diagnosis of MASLD (or NAFLD) or had evidence of hepatic steatosis on imaging with no competing chronic liver disease were followed from the time of MASLD ascertainment until the occurrence of a FIB-4 score at high-risk for advanced fibrosis (FIB-4 ≥ 2.67). Cox regression models were used to evaluate the relationship between objective measures of cardiometabolic risk with time to progression to a high-risk FIB-4.
Setting
This study used electronic health record (EHR) data from a general internal medicine primary care clinic between July 1, 2012 and December 31, 2021. Data were collected from the Enterprise Data Warehouse (EDW) and EPIC© Clarity database at the Medical University of South Carolina.
Study Sample
Patients receiving care from the primary care clinic with at least one visit during the study period and a diagnosis of MASLD (formerly NAFLD; International Classification of Diseases [ICD]-9/10: 571.8, K75.81, or K76.0), or a radiographic report of hepatic steatosis were considered for inclusion (15). Steatosis was identified from radiographic reports of all abdominal ultrasound, computed tomography, and magnetic resonance imaging during the study period. Natural language processing was applied to filter, search, and tabulate imaging report text to identify “hepatic steatosis (16).” Patients were also required to have qualifying inputs for a baseline FIB-4 calculation at the time of, or within the 1-year preceding, ascertainment of MASLD. Qualifying FIB-4 inputs included alanine (ALT) and aspartate (AST) aminotransferases (both < 350 IU/L) and a platelet count (Plt) within 6 months of the ALT and AST values. FIB-4 scores were calculated (FIB-4= [(Age x AST)/(Plt x √ALT)] and categorized by advanced fibrosis risk: low (FIB-4 < 1.3), indeterminate (1.3 ≤ FIB-4 < 2.67), and high-risk (FIB-4 ≥ 2.67) (10, 17, 18). Patients with a baseline FIB-4 ≥ 2.67 and those with a diagnosis (by ICD-9/10 code) of a competing chronic liver disease (i.e., non-MASLD), cirrhosis, hepatocellular carcinoma, or previous liver transplantation at baseline were excluded (Supplementary Table 1). Unfortunately, recorded alcohol use data in the EHR has low fidelity and is often incomplete, thus we had to rely on ICD-9/10 codes for competing alcohol-related liver disease. Additionally, patients with only one set of FIB-4 inputs were excluded.
Outcomes
Time to the occurrence of a FIB-4 score at high-risk for advanced fibrosis (FIB-4 ≥ 2.67) was the primary outcome of interest. Beginning with the baseline value, FIB-4 scores were calculated for each set of qualifying ALT, AST, and platelet values longitudinally. Each FIB-4 score is anchored to the date of the ALT and AST results and must be at least 6 months after the preceding FIB-4 value to ensure that all calculation inputs are only used once. Patients were followed from the baseline FIB-4 score until the occurrence of the first FIB-4 at high-risk for advanced fibrosis (FIB-4 ≥ 2.67). Patients without a high-risk FIB-4 during follow-up were followed until the end of the study period (December 31, 2021).
Primary Exposures
Objective measures of cardiometabolic risk factors, including body mass index (BMI), systolic blood pressure (SBP), hemoglobin A1c (HbA1c), low-density lipoprotein (LDL), triglycerides, thyroid stimulating hormone (TSH), estimated glomerular filtration rate (eGFR), and smoking status, were the primary exposures of interest. Each variable value identified in the EHR during follow-up was identified and patient-level means were calculated for each exposure. Using the mean values, each variable was categorized by a measure of cardiometabolic risk factor control. We used lower threshold categorical variables for BMI (≥ 30 kg/m2), SBP (≥ 140 mm Hg), HbA1c (≥ 7%), LDL (≥ 130 mg/dL), and triglycerides (≥ 150 mg/dL). We also had higher threshold categories for these risk variables with BMI (≥ 40 kg/m2), SBP (≥ 150 mm Hg), HbA1c (≥ 8%), LDL (≥ 160 mg/dL), and triglycerides (≥ 200 mg/dL). Only one threshold was used to categorize TSH (≥ 4.94 mIU/L) and eGFR (≤ 59 ml/min). Smoking status (Yes/No) was determined at the beginning of follow-up.
Other Covariates
Patient age was collected at the beginning of follow-up as a continuous variable. Sex (Female/Male), race (Non-white/White), and marital status (Married/Unmarried) were categorical variables. Patient comorbidities were identified using Elixhauser ICD-9/10 coding algorithms and included hypertension, diabetes mellitus, hyperlipidemia, cardiovascular disease (CVD), hypothyroidism, and chronic kidney disease (CKD) (19, 20). Given their relationship with cardiometabolic disease management and potential benefit in patients with chronic liver disease, statin prescriptions during follow-up were also identified within the EHR (21, 22). Statin prescriptions were treated as a categorical variable (Yes/No) based on receipt of at least one statin script during follow-up. We also identified prescriptions for glucagon-like peptide-1 receptor agonists (GLP-1α) and pioglitazone, as these medications may play a role in reducing steatosis and MASLD disease progression (23, 24).
Statistical Analysis
Cohort characteristics were described with proportions of categorical variables and the mean of continuous variables. Each available patient-level exposure variable result during follow-up was counted and the median number of each exposure variable in the cohort was calculated. The proportion of patients with mean exposure variables exceeding the low- and high-threshold values for control were calculated for the overall cohort and by the outcome of progressing to a high-risk FIB-4 score. Categorical exposure variables were compared using Chi square tests. Mean time from baseline to the end of follow-up was calculated for the cohort. Cox regression models were developed for the outcome of time to a high-risk FIB-4. Unadjusted models with each exposure variable were developed, as was a model adjusting for the other primary exposures. Additional adjusted Cox regression models were developed to control for statin prescriptions and demographic variables. We evaluated for significant interactions between statin, LDL, and triglyceride variables. Multicollinearity was assessed using variance inflation factor and models were tested for overall goodness of fit. SAS 9.4 (Cary, NC) was used for all statistical analyses.
Post Hoc Analysis
After developing and interpreting the pre-specified Cox regression models described above, we performed a sensitivity analysis including prescriptions for GLP-1α therapies in the fully adjusted model. Additionally, we performed a sensitivity analysis using a diagnosis of diabetes, identified by a composite of ICD-9/10 codes placed during follow-up using an Elixhauser algorithm, as a covariate in place of the objective hemoglobin A1c covariate (19, 20). Diabetes is a well-established risk factor for advanced fibrosis and cirrhosis in patents with MASLD, thus we developed unadjusted and adjusted Cox regression models to evaluate the association between diagnosed diabetes (with hemoglobin A1c removed) and the time to high advanced fibrosis risk (25).
Results
The MASLD cohort included 1,347 patients with an initial low- (n=859) or indeterminate-risk (n=488) FIB-4 score (Figure 1). The sample had a mean age of 54 years (SD ± 14), was 64% Female, and 40% non-White (Table 1). Of included patients, 78% had hypertension, 35% had diabetes, 32% had cardiovascular disease, and 17% had chronic kidney disease. Cohort patients had a median of 13 BMI (IQR: 5–31), 15 SBP (IQR: 6–33), 2 HbA1c (IQR: 1–5), 2 LDL (IQR: 1–3), 2 triglycerides (IQR: 1–4), 2 TSH (IQR: 1–2), and 6 eGFR (IQR: 2–13) values during the study period.
Figure 1.
Consort diagram for forming the cohort of patients with metabolic dysfunction-associated steatotic liver disease (MASLD) at low- and indeterminate-risk for advanced fibrosis.
Table 1.
Cohort characteristics overall and by baseline FIB-4 risk assessment.
| Overall | FIB-4 < 1.3 | 1.3 ≤ FIB-4 < 2.67 | |||
|---|---|---|---|---|---|
| Characteristics | n = 1,347 | n = 859 | n = 488 | p-value | |
| Demographics (%) | |||||
| Age, mean (SD) | 53.6 (14.2) | 49.6 (13.7) | 60.5 (12.4) | <0.001* | |
| Sex | <0.001† | ||||
| Female | 863 (64.1%) | 601 (70.0%) | 262 (53.7%) | ||
| Male | 484 (35.9%) | 258 (30.0%) | 226 (46.3%) | ||
| Race | 0.462† | ||||
| Non-White | 542 (40.2%) | 352 (41.0%) | 190 (38.9%) | ||
| White | 805 (59.8%) | 507 (59.0%) | 298 (61.1%) | ||
| Unmarried | 648 (48.1%) | 413 (48.1%) | 235 (48.2%) | 0.978† | |
| Exposure Variables, median counts (IQR) | |||||
| BMI | 13 (5–31) | 15 (6–32) | 11 (5–27) | 0.004‡ | |
| Systolic blood pressure | 15 (6–33) | 16 (6–34) | 13 (6–28) | 0.006‡ | |
| Hemoglobin A1c | 2 (1–5) | 2 (1–6) | 2 (1–4) | <0.001‡ | |
| LDL | 2 (1–3) | 2 (1–4) | 2 (1–3) | 0.024‡ | |
| Triglycerides | 2 (1–4) | 2 (1–4) | 2 (1–3) | 0.008‡ | |
| TSH | 2 (1–2) | 2 (1–3) | 1.5 (1–3) | 0.002‡ | |
| eGFR | 6 (2–13) | 6 (2–13) | 6 (2–12) | 0.913‡ | |
| Comorbidities (%) | |||||
| CVD | 434 (32.2%) | 276 (32.1%) | 158 (32.4%) | 0.926† | |
| Hypertension | 1,051 (78.0%) | 639 (74.4%) | 412 (84.4%) | <0.001† | |
| Diabetes mellitus | 467 (34.7%) | 310 (36.1%) | 157 (32.2%) | 0.147† | |
| Hyperlipidemia | 872 (64.7%) | 528 (61.5%) | 344 (70.5%) | <0.001† | |
| Smoking | 178 (13.2%) | 106 (12.3%) | 72 (14.8%) | 0.209† | |
| Hypothyroidism | 251 (18.6%) | 161 (18.7%) | 90 (18.4%) | 0.892† | |
| Chronic kidney disease | 229 (17.0%) | 140 (16.3%) | 89 (18.2%) | 0.362† | |
Two sample t test.
Chi square test.
Mann Whitney U test.
BMI=body mass index (kg/m2) CVD=cardiovascular disease. eGFR= estimated glomerular filtration rate. IQR=Interquartile range. LDL=low-density lipoprotein. SD=standard deviation. TSH=thyroid stimulating hormone.
Patients were followed for a mean 3.6 years (SD ± 2.5) with a median of 4 (IQR: 2–8) FIB-4 scores per patient, and of the cohort, 19% (n=258) had a follow-up FIB-4 score indicating a high-risk for advanced fibrosis. Of included patients, 60% had a mean BMI ≥ 30 kg/m2, 21% had a mean SBP ≥ 140 mm Hg, 24% had a mean HbA1c ≥ 7%, 30% had a mean eGFR ≤ 59 ml/min, and 13% smoked (Table 2). A higher proportion of patients that progressed to a high-risk FIB-4 had a mean SBP ≥ 150 mm Hg, a mean TSH ≥ 4.94 mIU/L, an EGFR ≤ 59 ml/min, and smoked at baseline compared to patients that did not have a FIB-4 ≥ 2.67 during follow-up. A lower proportion of patients with a high-risk FIB-4 during follow-up had a mean BMI ≥ 30 kg/m2 (50% vs. 63%, p<0.001) or mean BMI ≥ 40 kg/m2 (13% vs. 19%, p=0.024) compared to patients without a FIB-4 ≥ 2.67. During follow-up, 567 cohort patients received a prescription for statin therapy, 160 received a prescription for GLP-1α medication, and 5 received a script for pioglitazone.
Table 2.
Exposure variables for the overall cohort and by transition to a high-risk FIB-4.
| Overall | FIB-4 ≥ 2.67 | ||||
|---|---|---|---|---|---|
| Yes | No | p-value | |||
| Exposures | n = 1,347 | n = 258 | n = 1,089 | ||
| BMI | |||||
| ≥ 30 kg/m2 | 60.1% (810) | 49.6% (128) | 62.6% (682) | < 0.001* | |
| ≥ 40 kg/m2 | 18.0% (243) | 13.2% (34) | 19.2% (209) | 0.024* | |
| SBP | |||||
| ≥ 140 mm Hg | 20.8% (280) | 22.9% (59) | 20.3% (221) | 0.360* | |
| ≥ 150 mm Hg | 6.4% (86) | 9.3% (24) | 5.7% (62) | 0.033* | |
| Hb A1c | |||||
| ≥ 7% | 23.8% (320) | 24.0% (62) | 23.7% (258) | 0.908* | |
| ≥ 8% | 12.3% (165) | 12.8% (33) | 12.1% (132) | 0.768* | |
| LDL | |||||
| ≥ 130 mg/dL | 17.8% (240) | 10.1% (26) | 19.7% (214) | <0.001* | |
| ≥ 160 mg/dL | 5.8% (78) | 3.5% (9) | 6.3% (69) | 0.078* | |
| Triglycerides | |||||
| ≥ 150 mg/dL | 37.4% (504) | 32.6% (84) | 38.6% (420) | 0.073* | |
| ≥ 200 mg/dL | 20.2% (272) | 18.6% (48) | 20.6% (224) | 0.480* | |
| TSH | |||||
| ≥ 4.94 mIU/L | 3.3% (44) | 5.8% (15) | 2.7% (29) | 0.011* | |
| eGFR | |||||
| < 59 ml/min | 30.1% (405) | 53.1% (137) | 24.6% (268) | <0.001* | |
| Smoking | |||||
| Yes | 13.2% (178) | 18.6% (48) | 11.9% (130) | 0.005* | |
Chi square test.
BMI=body mass index (kg/m2) CVD=cardiovascular disease. eGFR= estimated glomerular filtration rate. FIB-4=Fibrosis-4 Index. Hb A1c=hemoglobin A1c. IQR=Interquartile range. LDL=low-density lipoprotein. SD=standard deviation. TSH=thyroid stimulating hormone.
In the unadjusted Cox regression model using the lower threshold cardiometabolic exposures, mean TSH ≥ 4.94 mIU/L (HR 1.87; 95%CI 1.11–3.14), mean eGFR ≤ 59 ml/min (HR 2.62; 95%CI 2.05–3.35), and smoking (HR 1.50; 95%CI 1.10–2.06) were associated with a greater hazard of progression to a high-risk for advanced fibrosis (Table 3). Mean BMI ≥ 30 kg/m2 (HR 0.62; 95%CI 0.49 – 0.79), mean LDL ≥ 130 mg/dL (HR 0.49; 95%CI 0.33–0.74), and mean triglycerides ≥ 150 mg/dL (HR 0.74; 95%CI 0.57–0.96) were associated with a lower hazard of progression to high-risk for advanced fibrosis. In the unadjusted model using the higher threshold cardiometabolic risk measures, mean BMI ≥ 40 kg/m2 was associated with a lower hazard (HR 0.68; 95%CI 0.47–0.97) of FIB-4 progression while a mean SBP ≥ 150 mm Hg was associated with an increased risk (HR 1.68; 95%CI 1.10–2.55) of having a FIB-4 ≥ 2.67 during follow-up.
Table 3.
Unadjusted and adjusted Cox regression models for the outcome of time to high-risk for advanced fibrosis (FIB-4 ≥ 2.67).
| Unadjusted | Adjusted* | |||
|---|---|---|---|---|
| Exposure | HR | 95% CI | HR | 95% CI |
| BMI ≥ 30 kg/m2 | 0.62 | 0.49 – 0.79 | 0.69 | 0.54 – 0.88 |
| SBP ≥ 140 mm Hg | 1.11 | 0.83 – 1.49 | 0.99 | 0.74 – 1.33 |
| A1c ≥ 7% | 0.92 | 0.69 – 1.23 | 0.82 | 0.61 – 1.11 |
| LDL ≥ 130 mg/dL | 0.49 | 0.33 – 0.74 | 0.54 | 0.36 – 0.82 |
| Trig ≥ 150 mg/dL | 0.74 | 0.57 – 0.96 | 0.72 | 0.55 – 0.94 |
| TSH ≥ 4.94 mIU/L | 1.87 | 1.11 – 3.14 | 1.69 | 1.00 – 2.86 |
| eGFR ≤ 59 ml/min | 2.62 | 2.05 – 3.35 | 2.67 | 2.07 – 3.44 |
| Smoking | 1.50 | 1.10 – 2.06 | 1.37 | 1.00 – 1.88 |
| Exposure | ||||
| BMI ≥ 40 kg/m2 | 0.68 | 0.47 – 0.97 | 0.70 | 0.49 – 1.00 |
| SBP ≥ 150 mm Hg | 1.68 | 1.10 – 2.55 | 1.57 | 1.03 – 2.40 |
| A1c ≥ 8% | 0.97 | 0.67 – 1.39 | 0.83 | 0.57 – 1.21 |
| LDL ≥ 160 mg/dL | 0.56 | 0.29 – 1.09 | 0.55 | 0.28 – 1.07 |
| Trig ≥ 200 mg/dL | 0.84 | 0.61 – 1.14 | 0.77 | 0.56 – 1.06 |
| TSH ≥ 4.94 mIU/L | 1.87 | 1.11 – 3.14 | 1.70 | 1.00 – 2.89 |
| eGFR ≤ 59 ml/min | 2.62 | 2.05 – 3.35 | 2.60 | 2.03 – 3.33 |
| Smoking | 1.50 | 1.10 – 2.06 | 1.44 | 1.05 – 1.97 |
Adjusted model adjusted for the other cardiometabolic risk factor exposures listed.
BMI=body mass index. eGFR=estimated glomerular filtration rate. LDL=low-density lipoprotein. SBP=systolic blood pressure. Trig=triglycerides. TSH=thyroid stimulating hormone.
After adjusting for the other cardiometabolic risk exposure variables, mean TSH ≥ 4.94 mIU/L, mean eGFR < 59 ml/min, and smoking status were associated with an increased hazard of progressing to high-risk for advanced fibrosis. Mean BMI ≥ 30 kg/m2, mean LDL ≥ 130 mg/dL, and mean triglycerides ≥ 150 mg/dL were associated with a lower risk of having a FIB-4 ≥ 2.67. In the adjusted model with the higher threshold cardiometabolic risks, mean SBP ≥ 150 mm Hg was associated with an increased hazard of progressing to a high-risk for advanced fibrosis along with TSH, eGFR, and smoking.
Adjusted Cox regression models were developed for low (Supplementary Table 2) and high threshold (Table 4) cardiometabolic risk exposures that included a variable for receipt of a statin prescription during follow-up. Using high threshold cardiometabolic risk exposures (Table 4), mean SBP ≥ 150 mm Hg (HR 1.65; 95%CI 1.08–2.52) and a mean eGFR ≤ 59 ml/min (HR 2.87; 95%CI 2.23–3.68) were associated with an increased risk of a FIB-4 ≥ 2.67, and mean BMI ≥ 40 kg/m2 (HR 0.69; 95%CI 0.49–0.99) and prescription of a statin (HR 0.50; 95%CI 0.39–0.65) were associated with a lower hazard of FIB-4 risk progression. After adjusting for demographic covariates, mean SBP ≥ 150 mm Hg (HR 1.57; 95%CI 1.02–2.41), mean eGFR ≤ 59 ml/min (HR 2.78; 95%CI 2.17–3.58), and male gender (HR 1.51; 95%CI 1.16–1.95) were associated with an increased risk a FIB-4 ≥ 2.67 while a statin prescription (HR 0.51; 95%CI 0.39–0.66) was associated with a lower hazard of progressing to a high-risk FIB-4. A sensitivity analysis with GLP-1α prescriptions added to the fully adjusted, high threshold cardiometabolic risk Cox regression model demonstrated a non-significant association between SBP≥150 mmHg and time to high-risk for advanced fibrosis (HR 1.46; 95%CI 0.95 – 2.24), and reduced hazard of fibrosis risk progression in patients with GLP-1α prescriptions (HR 0.37; 95%CI 0.20 – 0.69; Supplementary Table 3). All other variable relationships were similar to the primary analyses.
Table 4.
Cox regression models for the outcome of time to high-risk FIB-4 (FIB-4 ≥ 2.67) using the high threshold cardiometabolic risk factors after adjusting for statin prescriptions and demographic covariates.
| Adjusted for Statin Rx | Fully Adjusted | |||
|---|---|---|---|---|
| Exposure | HR | 95% CI | HR | 95% CI |
| BMI ≥ 40 kg/m2 | 0.69 | 0.48 – 0.99 | 0.71 | 0.49 – 1.03 |
| SBP ≥ 150 mm Hg | 1.65 | 1.08 – 2.52 | 1.57 | 1.02 – 2.41 |
| A1c ≥ 8% | 0.87 | 0.60 – 1.27 | 0.84 | 0.57 – 1.22 |
| LDL ≥ 160 mg/dL | 0.59 | 0.30 – 1.15 | 0.64 | 0.33 – 1.24 |
| Trig ≥ 200 mg/dL | 0.86 | 0.62 – 1.18 | 0.85 | 0.61 – 1.17 |
| TSH ≥ 4.94 mIU/L | 1.57 | 0.92 – 2.67 | 1.61 | 0.94 – 2.74 |
| eGFR ≤ 59 ml/min | 2.87 | 2.23 – 3.68 | 2.78 | 2.17 – 3.58 |
| Smoking | 1.35 | 0.98 – 1.85 | 1.22 | 0.88 – 1.69 |
| Statin Rx* | 0.50 | 0.39 – 0.65 | 0.51 | 0.39 – 0.66 |
| Non-White | 1.24 | 0.95 – 1.62 | ||
| Male | 1.51 | 1.16 – 1.95 | ||
| Unmarried | 1.17 | 0.90 – 1.52 | ||
Interactions between statin prescription and mean LDL ≥ 160 mg/dL (p=0.82) and Trig ≥ 200 mg/dL (p=0.84) were evaluated and found to be not statistically significant and were not included in the final models.
A1c=hemoglobin A1c. BMI=body mass index. eGFR=estimated glomerular filtration rate. LDL=low-density lipoprotein. Rx=prescription. SBP=systolic blood pressure. Trig=triglycerides. TSH=thyroid stimulating hormone.
In our post hoc analysis replacing measures of hemoglobin A1c with diabetes mellitus diagnoses by ICD-9/10 code, a diagnosis of diabetes during follow-up was associated with a higher hazard of advanced fibrosis risk progression (HR 1.31; 95%CI 1.02–1.68, Supplementary Table 4) in the unadjusted Cox regression models, but this relationship was not significant in the models adjusting for other cardiometabolic risk measures. Replacing objective measures of hemoglobin A1c control with diagnoses of diabetes led to a non-significant relationship between SBP≥150 mmHg and time to high-risk for advanced fibrosis in the fully adjusted Cox regression model (HR 1.52; 95%CI 0.99–2.33; Supplementary Table 5).
Discussion
In this retrospective cohort study of primary care patients with MASLD and low to indeterminate risk FIB-4 scores, 19% of patients progressed to a FIB-4 score at high-risk for advanced fibrosis (FIB-4 ≥ 2.67) during follow-up (mean 3.6 years). Uncontrolled hypertension (mean SBP ≥ 150 mm Hg) and kidney dysfunction (mean eGFR ≤ 59 ml/min) were associated with an increased risk of progression to a high-risk for advanced fibrosis when adjusting for other covariates in the primary analyses. Surprisingly, objective measures of hemoglobin A1c control were not associated with the hazard of progressing to high-risk for advanced fibrosis. Prescription of a statin therapy during follow-up was associated with a lower hazard of progressing to a high-risk FIB-4 score.
These results demonstrate a potential benefit for rigorous monitoring and treatment of cardiovascular risk factors for patients with MASLD that aligns with current chronic disease management goals in primary care (22, 26, 27). Optimizing blood pressure control and preserving kidney function could play roles in preventing patients from progressing to advanced fibrosis. Blood pressure management is a hallmark of primary care practice and a critical modifiable risk factor for preventing heart attack and stroke (26, 27). Cardiovascular disease is the leading cause of death in MASLD and blood pressure management in MASLD-affected patients may benefit from more frequent and intensive management resembling the care strategies used in patients with diabetes (27). Similarly, smoking demonstrated an association with FIB-4 risk progression and engaging patients in efforts to quit smoking can improve cardiovascular risk profiles and may positively contribute to liver health (28). However, smoking is often associated with alcohol use, thus this variable may also represent alcohol consumption which would certainly contribute to increased advanced fibrosis risk. Improved tobacco and alcohol exposure history is needed (29). Chronic kidney disease, and its progression, have been previously associated with MASLD and the use of traditional (i.e., angiotensin-converting enzyme inhibitors and angiotensin receptor blockers) and emerging (i.e., sodium glucose cotransporter-2 inhibitors) therapies may provide useful therapies for slowing the progression of advanced fibrosis in patients with MASLD (30, 31).
Uncontrolled hyperglycemia (mean HbA1c ≥ 8%) and BMI ≥ 40 kg/m2 were not significantly associated with time to a high-risk FIB-4 score. Though diabetes has been repeatedly recognized as a risk factor for advanced fibrosis and severe liver disease outcomes in patients with MASLD, elevations in an objective measure of glycemic control did not reveal this association (25). This finding may be due to limitations in the data (e.g., relatively few HbA1c’s per patient), the relatively short duration of follow-up, and potentially unmeasured confounding with other measures of cardiometabolic risk. We addressed this surprising finding by performing the post hoc analysis where we replaced the hemoglobin A1c covariate with a diabetes mellitus diagnosis (by ICD-9/10 code) covariate. Diagnoses of diabetes were associated with an increased hazard of progressing to a high-risk FIB-4 in the unadjusted analyses, but not after adjusting for other cardiometabolic risk measure covariates (25). The relationship of BMI with time to advanced fibrosis is more complicated. Though weight loss is a cornerstone of MASLD therapy and can prevent (and even reverse) advanced fibrosis, patients with MASLD with lower BMIs often fare no better than their counterparts with higher BMI values (11, 32). This finding also reinforces the concern that BMI may not be an optimal measure to assess obesity risk and further studies could focus on other measures including weight circumference, waist:hip ratio, waist:height ratio, or body composition analysis for body fat fraction (33–36). Glycemic control and weight loss are routinely addressed in primary care, and though their control likely has tremendous benefit in patients with MASLD, perhaps they are not the only cardiometabolic risk factors worth managing in this population.
The receipt of a statin prescription during follow-up was associated with a lower hazard of progressing to a high-risk FIB-4 (HR 0.51; 95%CI 0.39–0.66), when adjusting for the other cardiometabolic risk factors. Statins have been associated with improved outcomes in patients with chronic liver disease and cirrhosis and may play a key role in the future primary care management of MASLD (11, 21). This may also explain why elevated LDL and triglyceride values were associated with a lower risk of FIB-4 risk progression, as these elevations may have led to statin prescribing. However, there was no significant interaction between statin, LDL, and triglyceride variables. As of now, statins are critical tools in the prevention of cardiovascular disease and their use should not be withheld on account of mild elevations in serum liver chemistries, but MASLD does not factor into statin prescribing recommendations (22, 37). In the sensitivity analysis with GLP-1α prescriptions included as a covariate, receipt of a GLP-1α was associated with a lower risk of fibrosis progression, which further emphasizes the promise these medications hold for treating obesity, metabolic dysfunction, cardiovascular disease, and MASLD in the future. In this study, we cannot state with certainty how many patients prescribed these therapies were adherent due to limitations in our data.
We acknowledge limitations in this study. Most importantly, the primary outcome of interest is a FIB-4 serologic risk score and not a histologic outcome. High-risk FIB-4 values are not a perfect proxy for advanced fibrosis, but they have a high level of predictive performance for F3 and F4 fibrosis, are available longitudinally in primary care populations, and are tightly associated with severe liver disease outcomes including cirrhosis and hepatocellular carcinoma (17, 18, 38–41). Also, transition from low to high-risk for advanced fibrosis in MASLD takes time, and the relatively short duration of follow-up (mean 3.6 years) limits the interpretation of the results (8). This limitation also impacts the availability of measures of patient-level cardiometabolic risk factors, where some patients have only 1 available value for some measures. Fortunately, most patients in the sample have numerous BMI, systolic blood pressure, and eGFR values. Despite the short follow-up time, a relatively high proportion (19%) of patients progressed to a high-risk FIB-4 score. This finding could be due to our creation of a cohort with ICD-9/10 diagnoses of MASLD or with steatosis on abdominal imaging that may bias the sample to having more severe liver disease than the general population. Using steatosis for cohort inclusion may also result in including patients with other chronic liver diseases that were not completely evaluated clinically. Further, use of ICD-9/10 codes, with their limited sensitivity, to exclude other chronic liver disease and liver disease complications from the cohort at baseline might have missed competing liver disease etiologies that could potentially contribute to fibrosis risk change (19). Additionally, as an observational study, we cannot infer links of causality between measures of cardiometabolic risk and FIB-4 fibrosis risk progression. It is entirely possible that the relationships observed represent a simultaneous progression of worsening metabolic disease that impacts both cardiovascular health and the liver. Also, our race/ethnicity data is limited by incomplete ethnicity entries in the early years of our EHR data. Our future work will need to more comprehensively capture patient-reported ethnicity data given potential racial/ethnic disparities in MASLD-related outcomes. Lastly, these results come from a single center, which could threaten the generalizability of the results.
Conclusions
As the MASLD epidemic rages on, efforts to combat the disease will need to migrate to primary care. Detecting, and ideally preventing, advanced fibrosis plays a key role in the primary care management of MASLD. This study suggests that blood pressure control and preservation of kidney function may be therapeutic targets for care delivery in patients with MASLD at initially low-risk for advanced fibrosis.
Supplementary Material
Acknowledgments
All contributing authors have been recognized as such. We have no further acknowledgements.
Funding
National Institute of Diabetes and Digestive and Kidney Diseases (NIH/NIDDK K23DK118200 PI: Schreiner; R03DK129558 PI: Schreiner; P30DK123704 PI: Rockey). This project was also supported by the South Carolina Clinical & Translational Research Institute with an academic home at the Medical University of South Carolina CTSA National Center for Advancing Translational Sciences (NCATS) of the National Institutes of Health (NIH) under UL1 TR001450.
Footnotes
Competing interests
Authors JZ, WPM, DGK, JM, CB, PDM, and MG have nothing to declare. ADS has consulted for Novo Nordisk and Pfizer previously, but reports no conflicts with this study.
Ethics approval and consent to participate
This study (Pro00056541) involved human participants and was approved by the Institutional Review Board at the Medical University of South Carolina (MUSC).
Availability of data and materials
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality. The corresponding author (ADS) will on request detail the restrictions and any conditions under which access to some data may be provided.
References
- 1.Chalasani N, Younossi Z, Lavine JE, Charlton M, Cusi K, Rinella M, et al. The diagnosis and management of nonalcoholic fatty liver disease: Practice guidance from the American Association for the Study of Liver Diseases. Hepatology (Baltimore, Md). 2018;67(1):328–57. [DOI] [PubMed] [Google Scholar]
- 2.Rinella ME, Lazarus JV, Ratziu V, Francque SM, Sanyal AJ, Kanwal F, et al. A multisociety Delphi consensus statement on new fatty liver disease nomenclature. J Hepatol. 2023;79(6):1542–56. [DOI] [PubMed] [Google Scholar]
- 3.Rinella ME, Neuschwander-Tetri BA, Siddiqui MS, Abdelmalek MF, Caldwell S, Barb D, et al. AASLD Practice Guidance on the clinical assessment and management of nonalcoholic fatty liver disease. Hepatology (Baltimore, Md). 2023;77(5):1797–835. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Moon AM, Singal AG, Tapper EB. Contemporary Epidemiology of Chronic Liver Disease and Cirrhosis. Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association. 2019. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Kanwal F, Shubrook JH, Adams LA, Pfotenhauer K, Wai-Sun Wong V, Wright E, et al. Clinical Care Pathway for the Risk Stratification and Management of Patients With Nonalcoholic Fatty Liver Disease. Gastroenterology. 2021;161(5):1657–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Lazarus JV, Mark HE, Anstee QM, Arab JP, Batterham RL, Castera L, et al. Advancing the global public health agenda for NAFLD: a consensus statement. Nat Rev Gastroenterol Hepatol. 2021. [DOI] [PubMed] [Google Scholar]
- 7.Kanwal F, Shubrook JH, Younossi Z, Natarajan Y, Bugianesi E, Rinella ME, et al. Preparing for the NASH Epidemic: A Call to Action. Diabetes care. 2021. [DOI] [PubMed] [Google Scholar]
- 8.Sanyal AJ, Van Natta ML, Clark J, Neuschwander-Tetri BA, Diehl A, Dasarathy S, et al. Prospective Study of Outcomes in Adults with Nonalcoholic Fatty Liver Disease. N Engl J Med. 2021;385(17):1559–69. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9.Hagstrom H, Nasr P, Ekstedt M, Hammar U, Stal P, Hultcrantz R, et al. Fibrosis stage but not NASH predicts mortality and time to development of severe liver disease in biopsy-proven NAFLD. J Hepatol. 2017;67(6):1265–73. [DOI] [PubMed] [Google Scholar]
- 10.Vallet-Pichard A, Mallet V, Nalpas B, Verkarre V, Nalpas A, Dhalluin-Venier V, et al. FIB-4: an inexpensive and accurate marker of fibrosis in HCV infection. comparison with liver biopsy and fibrotest. Hepatology (Baltimore, Md). 2007;46(1):32–6. [DOI] [PubMed] [Google Scholar]
- 11.Cusi K, Isaacs S, Barb D, Basu R, Caprio S, Garvey WT, et al. American Association of Clinical Endocrinology Clinical Practice Guideline for the Diagnosis and Management of Nonalcoholic Fatty Liver Disease in Primary Care and Endocrinology Clinical Settings: Co-Sponsored by the American Association for the Study of Liver Diseases (AASLD). Endocr Pract. 2022;28(5):528–62. [DOI] [PubMed] [Google Scholar]
- 12.McPherson S, Hardy T, Dufour JF, Petta S, Romero-Gomez M, Allison M, et al. Age as a Confounding Factor for the Accurate Non-Invasive Diagnosis of Advanced NAFLD Fibrosis. Am J Gastroenterol. 2017;112(5):740–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Ciccarelli G, Di Giuseppe G, Cinti F, Moffa S, Mezza T, Giaccari A. Why do some glucose-lowering agents improve non-alcoholic fatty liver disease whereas others do not? A narrative review in search of a unifying hypothesis. Diabetes Metab Res Rev. 2023;39(7):e3668. [DOI] [PubMed] [Google Scholar]
- 14.Wong VWS, Zelber-Sagi S, Cusi K, Carrieri P, Wright E, Crespo J, et al. Management of NAFLD in primary care settings. Liver international : official journal of the International Association for the Study of the Liver. 2022;42(11):2377–89. [DOI] [PubMed] [Google Scholar]
- 15.Hagström H, Adams LA, Allen AM, Byrne CD, Chang Y, Grønbaek H, et al. Administrative Coding in Electronic Health Care Record-Based Research of NAFLD: An Expert Panel Consensus Statement. Hepatology (Baltimore, Md). 2021;74(1):474–82. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Nielsen EM, Anderson KP, Marsden J, Zhang J, Schreiner AD. Nonalcoholic Fatty Liver Disease Underdiagnosis in Primary Care: What Are We Missing? J Gen Intern Med. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.European Association for the Study of the L, List of panel m, Berzigotti A, Boursier J, Castera L, Cazzagon N, et al. Easl Clinical Practice Guidelines (Cpgs) On Non-Invasive Tests For Evaluation Of Liver Disease Severity And Prognosis- 2020 Update. J Hepatol. 2021. [Google Scholar]
- 18.Shah AG, Lydecker A, Murray K, Tetri BN, Contos MJ, Sanyal AJ, et al. Comparison of noninvasive markers of fibrosis in patients with nonalcoholic fatty liver disease. Clinical gastroenterology and hepatology : the official clinical practice journal of the American Gastroenterological Association. 2009;7(10):1104–12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Quan H, Li B, Saunders LD, Parsons GA, Nilsson CI, Alibhai A, et al. Assessing validity of ICD-9-CM and ICD-10 administrative data in recording clinical conditions in a unique dually coded database. Health Serv Res. 2008;43(4):1424–41. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 20.Quan H, Sundararajan V, Halfon P, Fong A, Burnand B, Luthi JC, et al. Coding algorithms for defining comorbidities in ICD-9-CM and ICD-10 administrative data. Med Care. 2005;43(11):1130–9. [DOI] [PubMed] [Google Scholar]
- 21.Vell MS, Loomba R, Krishnan A, Wangensteen KJ, Trebicka J, Creasy KT, et al. Association of Statin Use With Risk of Liver Disease, Hepatocellular Carcinoma, and Liver-Related Mortality. JAMA Netw Open. 2023;6(6):e2320222. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Force USPST, Mangione CM, Barry MJ, Nicholson WK, Cabana M, Chelmow D, et al. Statin Use for the Primary Prevention of Cardiovascular Disease in Adults: US Preventive Services Task Force Recommendation Statement. Jama. 2022;328(8):746–53. [DOI] [PubMed] [Google Scholar]
- 23.Newsome PN, Buchholtz K, Cusi K, Linder M, Okanoue T, Ratziu V, et al. A Placebo-Controlled Trial of Subcutaneous Semaglutide in Nonalcoholic Steatohepatitis. N Engl J Med. 2021;384(12):1113–24. [DOI] [PubMed] [Google Scholar]
- 24.Cusi K, Orsak B, Bril F, Lomonaco R, Hecht J, Ortiz-Lopez C, et al. Long-Term Pioglitazone Treatment for Patients With Nonalcoholic Steatohepatitis and Prediabetes or Type 2 Diabetes Mellitus: A Randomized Trial. Ann Intern Med. 2016;165(5):305–15. [DOI] [PubMed] [Google Scholar]
- 25.Huang DQ, Wilson LA, Behling C, Kleiner DE, Kowdley KV, Dasarathy S, et al. Fibrosis Progression Rate in Biopsy-Proven Nonalcoholic Fatty Liver Disease Among People With Diabetes Versus People Without Diabetes: A Multicenter Study. Gastroenterology. 2023;165(2):463–72 e5. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Whelton PK, Carey RM, Aronow WS, Casey DE Jr., Collins KJ, Dennison Himmelfarb C, et al. 2017 ACC/AHA/AAPA/ABC/ACPM/AGS/APhA/ASH/ASPC/NMA/PCNA Guideline for the Prevention, Detection, Evaluation, and Management of High Blood Pressure in Adults: A Report of the American College of Cardiology/American Heart Association Task Force on Clinical Practice Guidelines. Hypertension. 2018;71(6):e13–e115. [DOI] [PubMed] [Google Scholar]
- 27.Passarella P, Kiseleva TA, Valeeva FV, Gosmanov AR. Hypertension Management in Diabetes: 2018 Update. Diabetes Spectrum. 2018;31(3):218–24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Duncan MS, Freiberg MS, Greevy RA Jr, Kundu S, Vasan RS, Tindle HA. Association of Smoking Cessation With Subsequent Risk of Cardiovascular Disease. Jama. 2019;322(7):642–50. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 29.Lynch KL, Twesten JE, Stern A, Augustson EM. Level of Alcohol Consumption and Successful Smoking Cessation. Nicotine Tob Res. 2019;21(8):1058–64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Liang Y, Chen H, Liu Y, Hou X, Wei L, Bao Y, et al. Association of MAFLD With Diabetes, Chronic Kidney Disease, and Cardiovascular Disease: A 4.6-Year Cohort Study in China. J Clin Endocrinol Metab. 2022;107(1):88–97. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Rossing P, Caramori ML, Chan JCN, Heerspink HJL, Hurst C, Khunti K, et al. Executive summary of the KDIGO 2022 Clinical Practice Guideline for Diabetes Management in Chronic Kidney Disease: an update based on rapidly emerging new evidence. Kidney Int. 2022;102(5):990–9. [DOI] [PubMed] [Google Scholar]
- 32.Vilar-Gomez E, Martinez-Perez Y, Calzadilla-Bertot L, Torres-Gonzalez A, Gra-Oramas B, Gonzalez-Fabian L, et al. Weight Loss Through Lifestyle Modification Significantly Reduces Features of Nonalcoholic Steatohepatitis. Gastroenterology. 2015;149(2):367–78 e5; quiz e14–5. [DOI] [PubMed] [Google Scholar]
- 33.Pang Q, Zhang JY, Song SD, Qu K, Xu XS, Liu SS, et al. Central obesity and nonalcoholic fatty liver disease risk after adjusting for body mass index. World J Gastroenterol. 2015;21(5):1650–62. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Sheng G, Xie Q, Wang R, Hu C, Zhong M, Zou Y. Waist-to-height ratio and non-alcoholic fatty liver disease in adults. BMC Gastroenterol. 2021;21(1):239. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Xia T, Du M, Li H, Wang Y, Zha J, Wu T, et al. Association between Liver MRI Proton Density Fat Fraction and Liver Disease Risk. Radiology. 2023;309(1):e231007. [DOI] [PubMed] [Google Scholar]
- 36.Borges-Canha M, Neves JS, Silva MM, Mendonça F, Moreno T, Ribeiro S, et al. Waist-to-Hip Ratio and Inflammatory Parameters Are Associated with Risk of Non-Alcoholic Fatty Liver Disease in Patients with Morbid Obesity. Biomedicines. 2022;10(10). [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Athyros VG, Tziomalos K, Gossios TD, Griva T, Anagnostis P, Kargiotis K, et al. Safety and efficacy of long-term statin treatment for cardiovascular events in patients with coronary heart disease and abnormal liver tests in the Greek Atorvastatin and Coronary Heart Disease Evaluation (GREACE) Study: a post-hoc analysis. Lancet. 2010;376(9756):1916–22. [DOI] [PubMed] [Google Scholar]
- 38.Wong YJ, Urias E, Song MW, Goyal T, Tay WX, Han NX, et al. Combination of Fibrosis-4, liver-stiffness measurement, and Fibroscan-AST score to predict liver-related outcomes in nonalcoholic fatty liver disease. Hepatol Commun. 2023;7(10):e0244. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 39.Schreiner AD, Zhang J, Moran WP, Koch DG, Livingston S, Bays C, et al. Real-World Primary Care Data Comparing ALT and FIB-4 in Predicting Future Severe Liver Disease Outcomes. J Gen Intern Med. 2023;38(11):2453–60. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 40.Schreiner AD, Zhang J, Moran WP, Koch DG, Marsden J, Livingston S, et al. FIB-4 and incident severe liver outcomes in patients with undiagnosed chronic liver disease: A Fine-Gray competing risk analysis. Liver international : official journal of the International Association for the Study of the Liver. 2023;43(1):170–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 41.Schreiner AD, Zhang J, Durkalski-Mauldin V, Livingston S, Marsden J, Bian J, et al. Advanced Liver Fibrosis and the Metabolic Syndrome in a Primary Care Setting. Diabetes Metab Res Rev. 2021. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality. The corresponding author (ADS) will on request detail the restrictions and any conditions under which access to some data may be provided.