Hepatic Insulin Resistance Increases Risk of Gallstone Disease in Indigenous Americans in the Southwestern United States

Abstract

INTRODUCTION:

Animal models indicate that hepatic insulin resistance (IR) promotes cholesterol gallstone disease (GSD). We sought to determine whether hepatic and whole-body IR is associated with incident GSD.

METHODS:

At baseline, 450 Southwestern Indigenous American adults without GSD were included. Participants had a 2-step hyperinsulinemic-euglycemic clamp with glucose tracer at submaximal and maximal insulin stimulation (240 and 2,400 pmol/m²/min) for whole-body IR (M-low and M-high) and hepatic glucose production (HGP) before and during submaximal insulin infusion (HGP-basal and HGP-insulin). Incident GSD was identified during follow-up visits conducted at ∼2-year intervals. The associations of HGP (basal, insulin, and % suppression), M-low, and M-high with risk of GSD were assessed by Cox regression models adjusted for age, sex, body fat (%), glucose, and insulin.

RESULTS:

Sixty participants (13%) developed GSD (median follow-up: 11.6 years). Participants who developed GSD were of similar age and whole-body IR as those who did not (P's > 0.07) but were more likely to be female; have higher body fat, higher HGP-basal, and HGP-insulin; and lower % suppression of HGP (P's < 0.02). In separate adjusted models, higher HGP-insulin and lower % suppression of HGP were associated with increased risk for GSD (hazard ratio [HR] per SD: HR 1.38, 95% CI 1.12–1.69, P = 0.002; HR 1.41, 95% CI 1.16–1.72, P = 0.0007). HGP-basal, M-low, and M-high were not associated with GSD in adjusted models (P's > 0.22).

DISCUSSION:

Resistance to insulin suppression of HGP increases risk for GSD. Hepatic IR is a link between GSD and other conditions of the metabolic syndrome.

INTRODUCTION

Gallstone disease (GSD) is common and costly, and greatly contributes to morbidity worldwide, with a prevalence of approximately 15% in the United States and European populations (1). Although deaths directly due to complications of GSD are uncommon today (2), GSD is associated with increased mortality from cardiovascular disease and cancer. Thus, understanding the factors for the development of GSD is important (3).

Cholesterol stones are the most frequent types of gallstones (∼75%) including among Indigenous Americans (IAs) in the United States, a group with high prevalence rates for this disease (4,5). Compared with White populations, GSD is markedly higher in Indigenous (6) and Mexican American populations in the United States (7). In the presence of excess bile cholesterol, the bile becomes supersaturated with cholesterol. Under these conditions, cholesterol which is normally dissolved in micelles and vesicles precipitates from bile as solid monohydrate crystals (5). Importantly, cholesterol supersaturation has been shown to originate in hepatic bile, indicating mechanisms in the liver that promote these physicochemical abnormalities (8).

In addition to Indigenous and Mexican American heritage, risk factors for cholesterol gallstones include female sex, obesity, increasing age, genetic predisposition, rapid weight loss, use of certain medications (e.g., hormone replacement and oral contraceptives), and special conditions associated with gallbladder stasis (5). Patients with type 2 diabetes are at increased risk of GSD. IAs living in the Southwestern United States are at particularly high risk of developing both GSD and type 2 diabetes, suggesting pathways common to both diseases (9). Southwestern IAs (SIAs) have a high prevalence of obesity, and insulin resistance is an important pathophysiologic factor in the development of diabetes in this population (10). Compared with non-Hispanic Black and non-Hispanic White persons, SIAs demonstrate greater insulin resistance (11).

Given the co-occurrence of GSD, type 2 diabetes, and insulin resistance in IAs, Carey and Paigen hypothesized that insulin resistance may be contributing to lithogenic bile (12). However, human evidence linking insulin resistance and GSD has been limited. Elevated fasting insulin, a surrogate marker for whole-body insulin resistance (13), is associated with GSD in cross-sectional studies (14–17). In studies conducted in Chilean Hispanic (18), Turkish (19), and Korean populations, Homeostatic Model Assessment for Insulin Resistance (HOMA-IR), another surrogate measure of whole-body insulin resistance, is also associated with GSD (20).

Recent evidence from a liver-specific insulin receptor knockout (LIRKO) mice has indicated a direct role for hepatic insulin resistance (as opposed to whole-body insulin resistance) (21). LIRKO mice are unable to suppress hepatic glucose production (HGP) as measured with hyperinsulinemic-euglycemic clamp (HIEC) (22). These mice produce bile supersaturated with cholesterol and develop gallstones when fed a lithogenic diet (21). Whether hepatic insulin resistance is a risk factor in humans is unclear.

Prospective studies examining the potential relationship between GSD and insulin resistance using gold-standard measures of whole-body and liver-specific insulin resistance are lacking. This study sought to determine whether hepatic and whole-body insulin resistance is associated with the later development of GSD in a cohort of SIAs in the United States, measured with the HIEC combined with a glucose tracer to directly estimate both whole-body and liver-specific insulin resistance.

METHODS

Study design

This analysis included SIAs who were enrolled to a longitudinal cohort study in the Southwestern United States (NCT00339482) (9). Participants were invited to outpatient clinic visits every 2 years from the age of 5 years. GSD was identified by clinical history and chart review during these outpatient visits conducted by trained physicians (i.e., documented presence of gallstones by imaging or of a cholecystectomy). As adults, a subset of the participants in this study agreed to participate in an inpatient study with detailed metabolic phenotyping (NCT00340132) (10). Both studies were approved by the Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases. Written informed consent was obtained from all volunteers. All authors had access to the study data and reviewed and approved the final manuscript.

During the inpatient visit to the clinical research unit in Phoenix, Arizona (1982–2007), volunteers were screened with medical history, physical examination, and routine screening laboratory tests to ensure that they were healthy. No participants were taking medications that affect glucose metabolism. Women were nonpregnant. Study inclusion and exclusion criteria are shown in Supplementary Methods (see Supplementary Digital Content 1, http://links.lww.com/CTG/B194). After screening, participants were admitted and placed on a weight-maintaining diet (50% carbohydrate, 30% fat, and 20% protein) based on sex and weight (23). After at least 3 days on this diet, an oral glucose tolerance test (OGTT) was performed to confirm that participants did not have diabetes according to 2003 American Diabetes Association criteria (fasting plasma glucose <126 mg/dL and 2 hour-plasma glucose <200 mg/dL) (24). During this inpatient visit, phenotyping procedures included body composition measurements and 2-step HIEC with a glucose tracer to measure whole-body insulin sensitivity and HGP (basal and during insulin infusion). After discharge, participants were followed every 2 years to assess their GSD as part of the larger longitudinal cohort study described above (see Supplementary Figure 1, Supplementary Digital Content 2, http://links.lww.com/CTG/B195 for study timeline). For participants with a history of GSD, the first year in which GSD was documented was recorded. Participants with GSD before the HIEC were excluded.

Body composition

Body composition was measured by hydrodensitometry or dual-energy x-ray absorptiometry (DPX-L and Prodigy; GE/Lunar, Madison, WI) (25). Absorptiometric measurements were converged to comparable hydrodensitometry values as previously described (26,27).

Oral glucose tolerance test

Participants were given a 3-hour 75 g OGTT with plasma venous glucose measured at 0, 0.5, 1, 2, and 3 hours. Total glucose area under the curve (AUCs) for glucose concentrations during the OGTT were determined by the trapezoidal method. The HOMA-IR index (homeostasis model of insulin resistance) was calculated as (fasting glucose [mg/dL] × fasting insulin [mU/L]/405), with lower values indicating a higher degree of insulin sensitivity.

Two-step hyperinsulinemic-euglycemic clamp with glucose tracer

A 2-step HIEC was performed to measure the action of insulin at submaximal and max insulin concentrations, as previously described (10). In brief, after an overnight fast, a primed continuous intravenous insulin infusion was administered for 100 minutes at 240 pmol/m² body surface area/min (submaximal insulin stimulation), followed by a second 100 minutes at 2,400 pmol/m² body surface area/min (maximal stimulation). During both steps, a 20% dextrose solution was infused at varying rates to maintain a plasma glucose level of 100 mg/dL. The rate of glucose required to maintain euglycemia during hyperinsulinemic infusion is a measure of total insulin mediated glucose disposal, or insulin action (M), and was calculated for the past 40 minutes of each of the 2 insulin infusion rates (M-low and M-high) and corrected for the steady-state plasma glucose level. M-low was corrected for HGP during the submaximal insulin infusion. M-low and M-high were standardized to account for estimated metabolic body size calculated as fat-free mass +17.7 kg as previously described (28). HGP was determined at baseline before insulin infusion (HGP-basal) and at the end of the low-dose insulin infusion (HGP-insulin) using a primed (30 μCi), continuous (0.3 μCi/min) infusion of [3-H³]-glucose, calculated by the Steele equation (29). Suppression of HGP was calculated as the difference between HGP-basal and HGP-insulin, divided by HGP-basal, and expressed as a percentage. HGP during the HIEC (with a glucose tracer) provides an estimate of the suppressive action of insulin on glucose production: The higher the glucose production rate during the insulin stimulus (i.e., HGP-insulin), the higher the hepatic insulin resistance. Less suppression of basal glucose production by infused insulin indicates higher insulin resistance (30).

Analytical measures

All plasma glucose concentrations were determined by the glucose oxidase method (Beckman Instruments, Fullerton, CA), and plasma insulin concentrations were determined by the Herbert modification (31).

Statistical analyses

Statistical analyses were performed using SAS software (SAS Institute, Cary, NC). Baseline data of participants who later developed GSD were compared with those who did not develop GSD. Normally distributed data are reported as mean ± SD, whereas skewed data were reported as median and interquartile range. Independent sample t tests (continuous variables) or χ² tests (categorical variables) were used for group comparisons (GSD vs no GSD).

Cox proportional hazard models were used to evaluate the prospective association between insulin resistance measures (M-low, M-high, HGP-basal, HGP-insulin, and suppression of HGP) and GSD risk. Hazard ratios (HRs) are expressed for a 1 SD difference in the distribution of all continuous independent variables. For M-low, M-high, and % suppression of HGP, the inverse of the HRs was reported such that a higher value associated with a higher HR would indicate greater insulin resistance as a risk factor for GSD. Person time started at the inpatient visit (HIEC date) and extended to the year of GSD for those who developed disease or the year of last follow-up visit for those who did not. As we only collected year of onset, we used baseline year and year of onset to calculate time-to-event. Cox proportional hazards models were progressively adjusted models as follows: (i) unadjusted, (ii) adjusted for age and body fat (%), (iii) further adjusted for sex, (iv) further adjusted for glucose AUC, and (v) further adjusted for insulin concentrations corresponding to the relevant measurements (e.g., fasting insulin for HGP-basal, insulin during the low-dose infusion for HGP-insulin). Schoenfeld residuals were used to evaluate proportionality assumptions. A post hoc power analysis was conducted using a HR of 1.41 for the adjusted % suppression of HGP; calculated power was 0.76.

The follow-up period was shortened to 25 years to meet the proportionality assumption. Sensitivity analysis was conducted in a subgroup restricted to female patients only. Predicted cumulative incidence functions and their 95% confidence intervals were generated for the 10th and 90th percentiles of HGP-insulin and % suppression of HGP, assuming mean values for covariates including age, sex, body fat (%), glucose AUC, and insulin to visually assess the relationship between hepatic insulin resistance and GSD.

RESULTS

Baseline characteristics

Of the 648 participants with a HIEC measurement, 47 were excluded because of the presence of diabetes at baseline and 150 because of prior GSD. One participant was excluded because of an outlying value for insulin resistance value (>12.6 SD). Thus, 450 participants with HIEC without GSD and without diabetes were included at baseline (see Supplementary Figure 2, Supplementary Digital Content 3, http://links.lww.com/CTG/B196). Of the 450 participants with a median follow-up time of 11.6 years (interquartile range 7–15 years), 60 (13%) were diagnosed with GSD. Baseline characteristics are given in Table 1.

Table 1.

Baseline characteristics of participants who did (+) and did not (−) develop gallstone disease

Variable	Total	(+) GSD	(−) GSD
n (%)	450 (100)	60 (13)	390 (87)
Demographics
Age (yr)	26.4 ± 6.0	25.8 ± 6.1	26.5 ± 6.0
Sex, n (%)
Female	174 (39)	49 (11)a	125 (28)
Male	276 (61)	11 (2)	265 (59)
Body composition
Weight (kg)	92.8 ± 22.8	91.4 ± 19.9	93.0 ± 23.3
BMI (kg/m2)	33.3 ± 7.4	35.0 ± 7.1	33.0 ± 7.5
Body fat (%)	32.1 ± 8.4	37.2 ± 7.3a	31.3 ± 8.3
Fat mass (kg)	30.8 ± 13.3	34.8 ± 12.5b	30.2 ± 13.4
Fat-free mass (kg)	62.0 ± 12.8	56.6 ± 10.3a	62.8 ± 12.9
OGTT
Fasting plasma glucose (mg/dL)	88.1 ± 10.0	91.0 ± 10.9a	87.6 ± 9.7
2-hr plasma glucose (mg/dL)	119.6 ± 30.5	127.1 ± 28.6b	118.4 ± 30.7
Glucose regulation status, n (%)
NGR	316 (70)	34 (7)b	282 (63)
IGR	134 (30)	26 (6)	108 (24)
Glucose AUC (mg/dL × 180)	22,019 ± 4,110	23,019 ± 4,124b	21,863 ± 4,091
HIEC
HGP: basal (mg/kg EMBS/min)	1.91 ± 0.25	1.99 ± 0.27b	1.90 ± 0.24
HGP: insulin (mg/kg EMBS/min)	0.37 ± 0.42	0.50 ± 0.41b	0.35 ± 0.42
HGP: % suppression	81.0 ± 21.7	74.4 ± 22.0b	82.0 ± 21.5
M-low (mg/kg EMBS/min)	2.75 ± 1.16	2.49 ± 0.70	2.79 ± 1.21
M-high (mg/kg EMBS/min)	8.61 ± 2.19	8.33 ± 1.94	8.65 ± 2.22
Insulin: fasting (pmol/L)	226 ± 121	244 ± 145	223 ± 117
Insulin: low dose (pmol/L)	882 ± 293	874 ± 418	883 ± 270
Insulin: high dose (pmol/L)	15,420 ± 7,200	14,184 ± 5,467	15,606 ± 7,410
Follow-up
Follow-up time (yr)c	11 (7–15)	7 (4–13)	12 (8–16)

Values are expressed as mean ± SD or n (%) unless specified otherwise.

BMI, bod mass index; EMBS, estimated metabolic body size calculated as fat-free mass + 17.7 kg; glucose AUC, glucose area under the curve from a 3-hour OGTT; GSD, gallstone disease; HGP, hepatic glucose production; HIEC, hyperinsulinemic-euglycemic clamp; IGR, impaired glucose regulation; IQR, interquartile range; NGR, normal glucose regulation; OGTT, oral glucose tolerance test.

^aP < 0.01.

^bP < 0.05.

^cMedian (IQR).

Those with and without GSD had similar baseline age, weight, and BMI (all P's > 0.05). As expected, participants who developed GSD were more likely to be female (P < 0.0001) and to have higher body fat percentage (P < 0.0001), fat mass (P = 0.01), and lower fat-free mass (P = 0.0004). Fasting, 2-hour plasma glucose, and glucose AUC were higher in the participants who developed GSD (P = 0.006, P = 0.04, P = 0.04, respectively).

Those with and without GSD had similar whole-body glucose uptake (M-low, P = 0.07; M-high P = 0.30). However, participants who developed GSD had higher HGP during the basal state and during the submaximal insulin stimulation than those who did not progress (HGP-basal: P = 0.01; HGP-insulin: P = 0.01). Participants who developed GSD also had a decreased % suppression of HGP compared with the nonprogressors (P = 0.01).

Age, body fat (%), glucose AUC, and fasting insulin concentrations were correlated with M-low (r = −0.18, r = −0.44, r = −0.44, r = −0.78, respectively; all P < 0.001) and M-high (r = −0.11, r = −0.27, r = −0.40, r = −0.55, respectively; all P < 0.03). Body fat (%), glucose AUC, and fasting insulin concentrations were correlated with HGP-insulin (r = 0.28, r = 0.23, r = 0.45, respectively; all P < 0.0001) and % suppression of HGP (r = −0.27, r = −0.23, r = −0.45, respectively; all P < 0.0001). Female participants had higher HGP-basal (P = 0.001) and lower M-low (P = 0.02).

Hepatic glucose production and incident GSD

The HRs and 95% confidence intervals from Cox proportional hazards models evaluating the prospective association between baseline HGP and incident GSD are given in Table 2. In the unadjusted model, HGP-basal (HR 1.36 per SD, 95% CI 1.06–1.74, P = 0.02) was associated with increased risk of GSD (model 1). In a model adjusting for age and body fat (%), association between HGP-basal and GSD was no longer significant (HR 1.19, 95% CI 0.93–1.52, P = 0.18; model 2). Similarly, there was no relationship between HGP-basal and GSD in models further adjusted for sex (P = 0.22, model 3), glucose AUC (P = 0.23, model 4), and fasting plasma insulin (P = 0.22, model 5). As expected, female participants had higher risk of GSD in the full model (HR 5.3, P < 0.0001, model 5).

Table 2.

Hazard ratios and 95% confidence intervals of the association between hepatic glucose production and gallstone disease, all participants

	Model 1	Model 2	Model 3	Model 4	Model 5
HGP: basal	1.36 (1.06–1.74)a	1.19 (0.93–1.52)	1.17 (0.91–1.50)	1.17 (0.91–1.50)	1.17 (0.91–1.51)
Age	—	1.00 (0.74–1.34)	1.02 (0.76–1.38)	1.04 (0.76–1.41)	1.03 (0.76–1.41)
Body fat (%)	—	1.81 (1.35–2.42)b	1.16 (0.83–1.63)	1.17 (0.83–1.64)	1.16 (0.81–1.67)
Female sex	—	—	5.21 (2.44–11.1)b	5.3 (2.45–11.5)b	5.3 (2.45–11.5)b
Glucose AUC	—	—	—	0.98 (0.75–1.28)	0.98 (0.75–1.29)
Insulin	—	—	—	—	1.00 (0.77–1.29)
HGP: insulin	1.26 (1.05–1.50)a	1.25 (1.01–1.55)a	1.36 (1.10–1.67)b	1.37 (1.12–1.68)b	1.38 (1.12–1.69)b
Age	—	1.00 (0.74–1.34)	1.05 (0.77–1.43)	1.09 (0.80–1.50)	1.09 (0.80–1.50)
Body fat (%)	—	1.88 (1.40–2.52)b	1.15 (0.82–1.62)	1.16 (0.82–1.64)	1.18 (0.83–1.68)
Female sex	—	—	5.90 (2.74–12.7)b	6.37 (2.90–14.0)b	6.29 (2.86–13.8)b
Glucose AUC	—	—	—	0.90 (0.69–1.18)	0.91 (0.69–1.19)
Insulin	—	—	—	—	0.95 (0.72–1.25)
HGP: % suppression	1.24 (1.05–1.47)a	1.26 (1.03–1.55)a	1.39 (1.14–1.70)b	1.41 (1.15–1.71)b	1.41 (1.16–1.72)b
Age	—	1.00 (0.74–1.34)	1.06 (0.78–1.44)	1.10 (0.81–1.51)	1.10 (0.81–1.51)
Body fat (%)	—	1.91 (1.42–2.56)b	1.16 (0.82–1.64)	1.17 (0.83–1.65)	1.19 (0.84–1.70)
Female sex	—	—	6.18 (2.85–13.4)b	6.75 (3.03–15.3)b	6.66 (2.99–14.8)b
Glucose AUC	—	—	—	0.89 (0.68–1.17)	0.90 (0.69–1.18)
Insulin	—	—	—	—	0.94 (0.72–1.24)

AUC, area under the curve from the oral glucose tolerance test; HGP, hepatic glucose production.

All continuous variables were standardized (mean = 0, SD = 1), and the hazard ratios are reported per 1 SD. For % suppression of HGP, the inverses of the hazard ratio are reported.

^aP < 0.05.

^bP < 0.01.

In the unadjusted analysis, HGP during the insulin infusion was associated with increased risk of GSD (HR 1.26, 95% CI 1.05–1.50, P = 0.01, model 1). HGP-insulin remained associated with GSD in sequential models including the full model (HR 1.38, 95% CI 1.12–1.69, P = 0.002, model 5). As visualized in Figure 1a, higher HGP-insulin is associated with higher predicted cumulative incidence of GSD.

Figure 1.

Predicted cumulative incidence of gallstone disease with 95% confidence intervals at the 10th (closed circles) and 90th (open circles) percentiles of (a) hepatic glucose production—insulin and (b) hepatic glucose production—% suppression after adjustment for age, sex, body fat (%), glucose area under the curve, and plasma insulin.

In the unadjusted model, % suppression of HGP was associated with increased risk of GSD (HR 1.24, 95% CI 1.05–1.47, P = 0.013, model 1). In the full model adjusted age, body fat, sex, glucose AUC, and plasma insulin, the association between % suppression of HGP and GSD persisted (HR 1.41, 95% CI 1.16–1.72, P = 0.0007, model 5). As visualized in Figure 1b, decreased % suppression of HGP by insulin was associated with higher predicted cumulative incidence of GSD.

In sensitivity analyses restricted to the subgroup of female participants, the results were unchanged (see Supplementary Table 1, Supplementary Digital Content 4, http://links.lww.com/CTG/B197). Both HGP-insulin and % suppression of HGP remained associated with incident GSD in the fully adjusted model (HR 1.35, 95% CI 1.04–1.76, P = 0.03; HR 1.45, 95% CI 1.08–1.95, P = 0.01; model 5). Alternate analysis with logistic regression was similar to the Cox proportional hazards models, both HGP-insulin and % suppression of HGP were associated with GSD (see Supplementary Table 2, Supplementary Digital Content 5, http://links.lww.com/CTG/B198).

Baseline whole-body insulin resistance and incident GSD

Cox proportional hazards models evaluating the prospective association between whole-body insulin action (M-low, M-high) and GSD are given in Table 3. In the unadjusted and final models, M-low was not significantly associated with GSD, although the confidence intervals were wide (HR 2.88, 95% CI 0.85–9.80, P = 0.09, model 1; HR 2.31, 95% CI 0.40–13.5, P = 0.35, model 5). M-high was also not associated with risk of GSD (all P's > 0.05). HOMA-IR and fasting insulin were not associated with risk of GSD (see Supplementary Table 3, Supplementary Digital Content 6, http://links.lww.com/CTG/B199). Alternate analysis with logistic regression was similar to the Cox proportional hazards models, and both M-low and M-high were not significantly associated with GSD (see Supplementary Table 4, Supplementary Digital Content 7, http://links.lww.com/CTG/B200).

Table 3.

Hazard ratios and 95% confidence intervals of the association between M-low and M-high and gallstone disease, all participants

	Model 1	Model 2	Model 3	Model 4	Model 5
M-low	2.88 (0.85–9.80)	1.00 (0.28–3.57)	1.44 (0.36–5.78)	1.75 (0.37–8.20)	2.31 (0.40–13.5)
Age	—	0.98 (0.72–1.32)	1.00 (0.74–1.36)	1.03 (0.76–1.40)	1.03 (0.76–1.40)
Body fat (%)	—	1.90 (1.39–2.59)a	1.16 (0.80–1.68)	1.15 (0.79–1.67)	1.16 (0.80–1.68)
Female sex	—	—	5.29 (2.47–11.3)a	5.69 (2.59–12.5)a	5.78 (0.40–13.5)a
Glucose AUC	—	—	—	0.91 (0.68–1.22)	1.05 (0.77–1.42)
Insulin	—	—	—	—	0.92 (0.68–1.24)
M-high	1.12 (0.87–1.46)	0.99 (0.76–1.30)	1.02 (0.78–1.33)	1.08 (0.80–1.46)	1.10 (0.81–1.50)
Age	—	0.99 (0.74–1.32)	1.02 (0.76–1.38)	1.05 (0.78–1.42)	1.11 (0.81–1.50)
Body fat (%)	—	1.88 (1.41–2.50)a	1.19 (0.85–1.65)	1.18 (0.85–1.65)	1.25 (0.89–1.76)
Female sex	—	—	5.51 (2.61–11.7)a	5.94 (2.76–12.8)a	5.74 (2.67–12.3)a
Glucose AUC	—	—	—	0.89 (0.67–1.19)	0.88 (0.66–1.18)
Insulin	—	—	—	—	0.79 (0.56–1.12)

AUC, area under the curve from the oral glucose tolerance test.

All continuous variables were standardized (mean = 0, SD = 1), and the inverse of the hazard ratios is reported per one 95% confidence interval.

^aP < 0.01.

DISCUSSION

This study investigated the link between GSD and insulin resistance using gold-standard measures of whole-body and liver-specific insulin resistance in a prospective cohort of IAs living in the Southwestern United States. Using direct measures of HGP and whole-body insulin resistance, increased absolute HGP during insulin infusion and decreased capacity to suppress HGP were associated with development of GSD, whereas whole-body insulin resistance was not associated. The associations between GSD and measures of hepatic insulin resistance were independent of important risk factors of GSD including increased age, female sex, obesity (body fat percentage), and plasma glucose concentrations. In these models, female sex remained a strong predictor of GSD. These findings implicate hepatic insulin resistance as a link between GSD and other diseases of metabolic syndrome.

LIRKO mice have diminished capacity to suppress HGP (22). Biddinger et al (21) demonstrated that this mouse model is at high risk of developing cholesterol gallstones due to a lithogenic bile profile, a profile resembling that found in SIAs (8). This study supports hepatic insulin resistance as a crucial link to human susceptibility to GSD. The LIRKO mouse model has severe glucose intolerance, defective insulin clearance (predominantly liver mediated), hyperinsulinemia, and decreased serum albumin (an insulin regulated gene in the liver) (22). The population of SIAs has high rates of both GSD and type 2 diabetes with the latter previously linked to hyperinsulinemia (32), lower insulin clearance (33), and decreased albumin (34), suggesting that hepatic insulin resistance may be contributing to a broader phenotypic presentation that includes gallstones.

The molecular mechanisms underlying the link between hepatic insulin resistance and gallstones in human populations including SIAs are unclear. In the LIRKO mouse model, Akt signaling is reduced and FoxO1 phosphorylation is impaired in the liver, leading to increased expression of ABCG5/ABCG8 (21). In other human populations including Chilean with IAs ancestry (35), single-nucleotide polymorphisms in ABCG5/ABCG8 are linked with GSD (36). Whether hepatic insulin resistance and GSD in SIAs have a genetic basis has not yet been investigated. Furthermore, there may be a role for altered gut microbiota (e.g., Desulfovibrionales) as these microbial alterations are also linked to states of increased hepatic insulin resistance such as obesity and metabolic syndrome (37,38).

A strength of this study was use of 2 insulin infusion rates within the gold-standard HIEC to measure whole-body insulin resistance. Since insulin has concentration-dependent saturable actions to increase whole-body glucose uptake, the submaximal and maximal stimulation (M-low and M-high) reduces the likelihood of missing potential differences in insulin action among insulin-resistant participants (39). Unlike hepatic insulin resistance, we found the whole-body insulin resistance did not predict GSD. Whole-body insulin resistance is a strong predictor of diabetes in this population (10) and others (5). M reflects that whole-body glucose uptake and glycemia are determined by uptake in other insulin-sensitive tissues besides the liver. Prior evidence indicates that muscle, which is found in larger amounts in the body, is more important than liver in determining glucose concentrations (40). This indicates that abnormalities in nonliver tissues are less relevant to the development of gallstones. That HGP during insulin suppression was associated with GSD may simply reflect the known relevance of the liver in producing cholesterol supersaturated bile.

The use of an isotopic glucose tracer allowed for measurement of HGP and was another strength of our study. First, it allowed a direct measurement of hepatic insulin resistance as opposed to using indices or biochemical markers that serve as surrogate measures. Second, simultaneous measurement of HGP improves estimates of whole-body insulin resistance. Although the exogenous glucose infusion rate in the HIEC is sometimes used as measure of whole-body insulin resistance, it does not account for the portion of the glucose rate of appearance produced endogenously.

Other strengths of the study include a prospective study design with sufficiently long follow-up time and a well-characterized cohort with adjustment for important risk factors for GSD including age, female sex, and obesity. Body fat percentage was used as a measure of adiposity with gold standard methods. Body fat percentage may be a more accurate measure of obesity compared with body mass index, which relies on only height and weight in its calculation.

The findings should be interpreted in the context of some limitations. First, cases of GSD relied on history and medical chart review. While routine surveillance gallbladder imaging (e.g., with ultrasound) may have detected more cases including asymptomatic gallstones, the gallstones in this study are more likely to be symptomatic cases and thus more clinically relevant. In addition, similar definitions and methods to find cases have been used to identify novel genetic loci associated with GSD (41). Although it is uncertain how inclusion of asymptomatic gallstones among the cases in this study would have changed the results, it is possible that this may increase power to detect an association between whole-body insulin resistance and gallstones as both asymptomatic and symptomatic gallstones share common pathogenic pathways (e.g., cholesterol supersaturated bile). Second, subclinical symptoms before diagnosis may have occurred in some participants before the year of clinical diagnosis. Thus, there may be some imprecision in the time of GSD onset. Third, there may be differences in the lithogenic potential of individuals' habitual diets that are difficult to quantify. In LIRKO mice, gallstones are most evident when these mice are fed a lithogenic diet (21). Data on habitual food intake by the study participants are unavailable. In general, dietary studies in this population indicate modern-day diets much higher in fat compared with traditional IA diets (12). Finally, all participants were IAs, a high-risk group for developing gallstones. SIAs are more insulin resistant than White and Black groups (11). Thus, results may not be fully generalizable to other populations. However, pathophysiological factors of GSD such as cholesterol supersaturated bile are common to both these SIAs (8) and other populations alike (7). In addition, the importance of insulin resistance in type 2 diabetes was subsequently replicated in other populations.

In summary, hepatic insulin resistance was associated with the development of gallstones in humans and provides a crucial link between susceptibility to gallstones and metabolic syndrome. Insulin resistance within the liver may have a specific role in the development of GSD and may serve as a therapeutic target for prevention of the disease.

CONFLICTS OF INTEREST

Guarantor of the article: Douglas C. Chang, MD.

Specific author contributions: B.N.A.: conceptualization, methodology, formal analysis, investigation, writing—original draft, visualization. E.J.S., T.C.D.B., R.L.H., and H.C.L.: conceptualization, methodology, formal analysis, investigation, writing—review and editing. J.K. and D.C.C.: conceptualization, methodology, formal analysis, investigation, writing—review and editing, visualization, supervision.

Financial support: This research was supported by the Intramural Research Program of the National Institutes of Health, National Institute of Diabetes and Digestive and Kidney Diseases.

Potential competing interests: None to report.

Clinical trial registration: ClinicalTrials.gov identifiers: NCT00339482, NCT00340132.

Data sharing statement: The deidentified datasets generated and/or analyzed during this study will be made available from the corresponding author upon reasonable request pending application and approval. Proposals for access should be sent to changdc@mail.nih.gov between 3 months and 3 years of publication.

Study Highlights.

WHAT IS KNOWN

• ✓ Animal models suggest a role for hepatic insulin resistance in the pathogenesis of gallstone disease.

• ✓ Human evidence linking insulin resistance to gallstones is limited.

WHAT IS NEW HERE

• ✓ Hepatic insulin resistance is associated with increased risk for development of gallstones.

• ✓ Hepatic insulin resistance is a potential target for prevention of gallstones and is a link to the metabolic syndrome.

Supplementary Material

ct9-15-e00763-s001.docx ^{(14KB, docx)}

ct9-15-e00763-s002.pdf ^{(116.8KB, pdf)}

ct9-15-e00763-s003.pdf ^{(27.4KB, pdf)}

ct9-15-e00763-s004.docx ^{(18.2KB, docx)}

ct9-15-e00763-s005.docx ^{(16.1KB, docx)}

ct9-15-e00763-s006.docx ^{(16.2KB, docx)}

ct9-15-e00763-s007.docx ^{(15.4KB, docx)}