|Year : 2020 | Volume
| Issue : 1 | Page : 55-61
Insulin resistance as a noninvasive predictor of esophageal varices in hepatitis C virus cirrhotic patients
Atef A Ali1, Belal A Montaser2, Heba G Abdel-Fattah1, Ahmed R El-Gazzarah1
1 Department of Tropical Medicine, Faculty of Medicine, Menoufia University, Menoufia, Egypt
2 Department of Clinical Pathology, Faculty of Medicine, Menoufia University, Menoufia, Egypt
|Date of Submission||18-Feb-2019|
|Date of Decision||26-Mar-2019|
|Date of Acceptance||30-Mar-2019|
|Date of Web Publication||25-Mar-2020|
Heba G Abdel-Fattah
Shebin Al Kom, Menoufia 32511
Source of Support: None, Conflict of Interest: None
The aim was to evaluate sensitivity and specificity of insulin resistance (IR) as a noninvasive predictor of esophageal varices (EV) in hepatitis C virus (HCV) cirrhotic patients.
Variceal bleeding due to portal hypertension is associated with a high probability of circulatory dysfunction and even death. However, routine endoscopy is an invasive maneuver which consumes effort, time, and money. IR was studied as an early noninvasive predictor of EV.
Patients and methods
Eighty cirrhotic patients were included in this prospective case–control study and 20 nondiabetic nonhepatic patients served as the control group. Patients were recruited from the Gastrointestinal Endoscopy Unit of Tropical Medicine Department, Menoufia University Hospitals from January 2017 to March 2018. IR was calculated by the homeostasis model assessment (HOMA-IR)=fasting insulin (μU/ml)×fasting glucose (mmol/l)/22.5.
HOMA-IR showed a high statistically significant correlation with the presence and grade of EV in HCV cirrhotic patients (P < 0.001). In comparison with others noninvasive predictors, HOMA-IR gave the highest sensitivity at a cutoff value of 4.41. Next, the midclavicular liver span/albumin ratio at a cutoff value of 3.51 was followed by the portal vein diameter at a cutoff value of 13 mm. The least sensitive predictor was the platelet count/splenic bipolar diameter ratio at a cutoff value of 1414.
IR estimated by HOMA-IR can provide sensitive information for determination of the presence and grade of EV in HCV cirrhotic patients regardless of their Child–Turcotte–Pugh classification.
Keywords: cirrhosis, hepatitis C virus, insulin resistance, portal hypertension, varices
|How to cite this article:|
Ali AA, Montaser BA, Abdel-Fattah HG, El-Gazzarah AR. Insulin resistance as a noninvasive predictor of esophageal varices in hepatitis C virus cirrhotic patients. Menoufia Med J 2020;33:55-61
|How to cite this URL:|
Ali AA, Montaser BA, Abdel-Fattah HG, El-Gazzarah AR. Insulin resistance as a noninvasive predictor of esophageal varices in hepatitis C virus cirrhotic patients. Menoufia Med J [serial online] 2020 [cited 2020 Mar 30];33:55-61. Available from: http://www.mmj.eg.net/text.asp?2020/33/1/55/281316
| Introduction|| |
Management of esophageal varices (EV) receives exceptional attention in cirrhotic patients due to the high probability of circulatory dysfunction and even death in case of variceal bleeding. In newly diagnosed cirrhotic patients, the prevalence of EV is ∼60–80% and the 1-year rate of first variceal bleeding is ∼5% for small EV and 15% for large EV.
Determination of the presence of EV by upper digestive endoscopy is therefore mandatory in patients with cirrhosis at diagnosis. For long-term follow-up, guidelines recommend monitoring of cirrhotic patients by routine endoscopy for the detection of the development of EV and to initiate prophylactic measures to prevent the bleeding of EV when they become large. Endoscopy is, however, a costly, invasive, and time-consuming procedure.
As noninvasive indicators of varices are desired to reduce the need for screening endoscopy in all patients with cirrhosis, many studies have tried to determine whether clinical or laboratory nonendoscopic parameters could predict the presence of large EV.
Researches about the interference between an impaired portal circulation and the glucose metabolism are limited, therefore several recent researches have studied the relationship between insulin resistance (IR) and hepatic hemodynamics as more accurate predictors for portal hypertension.
The relationship between portal hemodynamics and IR can be explained by two hypotheses. One is the diversion of blood flow away from the liver via collaterals leading to abnormal insulin vascular distribution and consequentially IR. Some recent studies have reported that the occlusion of shunts by balloon-occluded retrograde transvenous obliteration has increased portal venous inflow, resulting in enhanced hepatic insulin clearance and improved IR-related hyperinsulinemia in portal hypertensive patients.
However, Erice et al. have suggested that once hepatic venous pressure gradient (10 mmHg) [clinically significant PH (CSPH)] had developed, secondary hyperinsulinemia may develop due to two different mechanisms: either pancreatic hypersecretory state, as suggested by the moderately higher C-peptide levels in patients with IR, or due to reduction of insulin clearance as a consequence of portal-systemic shunting but not due to more intrinsic hepatocellular or endothelial pathology.
Many studies have concluded that CSPH had significant correlation with higher IR, more ascites, the presence of EV, and worse hepatocellular function than patients without CSPH.
The second hypothesis correlating IR and portal hemodynamics is the role of IR as a trigger for collateral development. Some studies considered IR as a cause rather than a result of steatosis and fibrosis in genotype 1 hepatitis C virus (HCV) patients, suggesting that increasing insulin level in the circulation could be a risk factor for fibrosis through IR-induced steatosis.
There is a strong relationship between HCV-mediated IR and type 2 diabetes mellitus development. Nearly 47 million patients may have HCV-associated diabetes mellitus. In HCV patients, the prevalence of hepatic steatosis is about 2–3-fold higher than the prevalence of steatosis due to other etiologies. IR enhances hepatic fibrosis and its progression to cirrhosis.
Because portal hypertension has multifactorial causes, it is difficult to state whether IR is a consequence or a trigger in portal hypertension. The substantial role of IR in the pathogenesis of portal hypertension remains to be elucidated. Many studies have discussed IR in the noninvasive prediction of portal hypertension. In conclusion, they reported that IR, regardless of the presence of diabetes, significantly predicts the presence of EV, in participants with HCV-related cirrhosis in order to reduce the need for endoscopic procedures.
On the basis of the above hypotheses about the correlation between IR-induced and HCV-induced cirrhosis, the present study investigates the sensitivity and specificity of IR as a noninvasive predictor of EV in exchange for endoscopy in HCV-cirrhotic patients.
| Patients and Methods|| |
This prospective case–control study included 100 participants, chosen according to the inclusion and exclusion criteria. The patients were recruited from the Gastrointestinal Endoscopy Unit of Tropical Medicine Department, Menoufia University Hospitals from January 2017 to March 2018 after approval from the Research and Ethics Committee of Faculty of Medicine, Menoufia University.
All participants included in the study signed an informed written consent. Data were collected in a preorganized data sheet (Case Sheet) by the researcher from patients fulfilling the inclusion and exclusion criteria.
Patients and controls were classified into the following groups:
- Group I: consisted of 80 patients diagnosed to have EV by endogastroduodenoscopy (EGD) and were proved to have HCV-induced cirrhosis by clinical criteria (history and physical examination) in addition to laboratory and imaging findings. The patients were classified into four subgroups according to the Paquet grading system of EV:
- Group IA: 20 patients with grade I EV
- Group IB: 20 patients with grade II EV
- Group IC: 20 patients with grade III EV
- Group ID: 20 patients with grade IV EV.
- Group II: 20 nondiabetic nonhepatic patients who have undergone diagnostic EGD (six patients for dyspepsia unresponsive to medical therapy, four patients for persistent gastroesophageal reflux symptoms, five patients for dysphagia, and five patients for occult gastrointestinal bleeding) showed no signs of portal hypertension as a control.
Patients with other causes of liver disease or mixed etiology, diabetic patients, patients with portal vein thrombosis, patients with any malignancies or any endocrinal disorders, and those with BMI greater than or equal to 30 kg/m2 were excluded from the study.
All participants were subjected to thorough history taking, clinical general and local examination, and laboratory investigations in the form of complete blood counts, liver profile, kidney function tests, fasting plasma glucose concentration, and viral markers (HCV antibodies, hepatitis B surface antigen and antibodies for human immunodeficiency) by enzyme-linked immunosorbent assay.
Fasting insulin (IU/ml) immunoenzymometric assay: was conducted using kits manufactured by Monobind (AccuBind ELISA Microwells; Monobind, Lake Forest, California, USA). In this procedure, immobilization takes place during the assay at the surface of a microplate well through the interaction of streptavidin coated on the well and exogenously added biotinylated monoclonal insulin antibody. Upon mixing the monoclonal biotinylated antibody, the enzyme-labeled antibody and a serum containing the native antigen react to form a soluble sandwich complex. Simultaneously, the complex is deposited on the wall through the high-affinity reaction of streptavidin and biotinylated antibody. After equilibrium, the antibody-bound fraction is separated from the unbound antigen. The enzyme activity of the antibody-bound fraction is proportional to the native antigen concentration. By utilizing several different serum antigen values, a dose–response curve can be generated from which the antigen concentration of an unknown can be ascertained.
Calculation of IR by homeostasis model assessment is according to the equation (HOMA-IR)=fasting insulin (μU/ml)×fasting glucose (mmol/l)/22.5.
Pelviabdominal ultrasound was performed for all participants to show the features of liver cirrhosis, measure portal vein diameter, measure the bipolar diameter of the spleen, and to detect perisplenic collaterals.
The patients were subjected to EGD within a maximum of 2 weeks of investigation of laboratory parameters. Endoscopy was performed by an expert endoscopist using a flexible video gastroscope (Olympus EVIS-200; Olympus, Shinjuku-ku, Tokyo, Japan) and variceal grade was evaluated according to Paquet and Palmer and Brick criteria:
- Grade 0: absent EV
- Grade 1: microvessels that sketch varicose strings located in the esophagogastric transition or in the distal esophagus
- Grade 2: one or two fine caliber varices (<3 mm in diameter) located in the distal esophagus
- Grade 3: medium caliber varices (3–6 mm in diameter) or varices up to 3 mm that may reach up to the medium third of the esophagus
- Grade 4: thick caliber varices (>6 mm in diameter), in any part of the esophagus.
Data were collected, coded and then entered into a spreadsheet using Microsoft Excel for Windows Office 2010 (One Microsoft Way, Redmond, Washington, USA.). Data were statistically analyzed using the Statistical Package of the Social Sciences SPSS 20: on IBM compatible computer (Armonk, New York, United States). Quantitative data were expressed as the mean ± SD or SE. SE = SD/square root of patients' number which was used in case of big SD; data were analyzed by independent sample, paired t-test and one-way analysis of variance followed by Tukey's test whenever single-step multiple comparisons were needed to maintain the integrity of data, while qualitative data were expressed as number and percentage and were analyzed by χ2) test. The correlation was done using Pearson's correlation test. The receiver operating characteristic curve and 95% confidence interval were performed to determine the cutoff values for the studied biomarkers. Sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV) were determined. P value was considered significant if less than 0.05 and highly significant if less than 0.001.
| Results|| |
The study was conducted on 100 participants: 51 (51%) of them were men and 49 (49%) of them were women with a mean age of 47.350 ± 5.603 years in group IA, 43.100 ± 6.619 years in group IB, 47.150 ± 8.190 years in group IC, 48.650 ± 6.055 years in group ID, and 44.950 ± 7.891 years in group II with a statistically nonsignificant difference between the studied groups regarding age and sex distribution.
Twenty-nine patients (29%) had ascites (P = 0.012); 18 patients (18%) had jaundice (P = 0.001); 33 patients (33%) had a previous history of variceal bleeding (P = 0.006); 29 patients (29%) had previous episode/s of hepatic encephalopathy (P = 0.018), and 16 patients (16%) had petechiae and/or ecchymosis (P = 0.006).
The mean fasting blood glucose value was101.850 ± 15.826 mg/dl in group IA, 107.500 ± 14.211 mg/dl in group IB, 103.450 ± 14.537 mg/dl in group IC, 104.500 ± 13.169 mg/dl in group ID, and 97.700 ± 12.620 mg/dl in group II with a statistically nonsignificant difference between the studied groups.
There was a high statistically significant difference between the studied groups regarding fasting blood insulin levels (P < 0.001) as shown in [Table 1].
By calculation of IR by homeostasis model assessment (HOMA-IR), there was high statistically significant difference between the studied groups regarding HOMA-IR index [Table 2].
|Table 2: Homeostasis model assessment insulin resistance index of the studied groups|
Click here to view
Correlation between HOMA-IR and demographic, laboratory and ultrasonographic data of all cases (n = 100) is demonstrated in [Table 3].
|Table 3: Correlation between homeostasis model assessment insulin resistance and demographic, laboratory, and ultrasonographic data of all cases (n=100)|
Click here to view
Receiver operating characteristic curve analysis was performed to assess the diagnostic performance of HOMA-IR between group II (control group) and each of group IA (grade I EV), group IB (grade II EV), group IC (grade III EV) and group ID (grade IV EV). Analysis showed that the best cutoff value of HOMA-IR in group IA was greater than 2.1 with a diagnostic sensitivity of 85%, a specificity of 65%, a PPV of 70.8%, and a NPV of 81.2% and the accuracy was 75.6% [Figure 1].
|Figure 1: Receiver operating characteristic curve for sensitivity and specificity of homeostasis model assessment insulin resistance score between: (a) groups IA and II, (b) groups IB and II, (c) groups IC and II, (d) groups ID and II.|
Click here to view
Correlation of HOMA-IR levels between group II and group IB showed that at a cutoff level greater than 4.41, the sensitivity was 90%, specificity 90%, PPV 90%, NPV 90%, and the accuracy was 95.3% [Figure 1].
Correlation of HOMA-IR levels between group II and each of group IC and group ID EV patients showed that at a cutoff level greater than 5.9, the sensitivity was 100%, specificity 100%, PPV 100%, NPV 100%, and accuracy 100% [Figure 1].
On comparing IR to other noninvasive predictors, HOMA-IR score gave the highest accuracy, then the mid-clavicular liver span/albumin ratio, followed by portal vein diameter and the least accurate was platelet count/splenic bipolar diameter ratio as shown in [Table 4].
|Table 4: Comparison between homeostasis model assessment insulin resistance, platelet count/splenic bipolar diameter ratio, mid-clavicular liver span/albumin concentration ratio, and portal vein diameter in predicting EV in all cases (n=100)|
Click here to view
| Discussion|| |
The current study shows a statistically significant correlation between HOMA-IR and each of mid-clavicular liver span, albumin level, bilirubin level, platelet count, portal vein diameter, and spleen bipolar diameter (P < 0.001). This can recommend considering HOMA-IR along with high portal vein diameter, splenomegaly, and low platelet count as a predictor of EV presence.
There is also a significant correlation between HOMA-IR and grade of varices. The best cutoff value was greater than 2.1 for grade I EV and greater than 4.41 for grade II EV, while HOMA-IR greater than 5.9 is suggestive of grade III and IV EV. On the basis of these data, HOMA-IR can predict the presence of EV and also the possibility of a patient needing therapeutic endoscopic modalities.
These results are consistent with Cammà et al., who concluded that the best cutoff value of HOMA-IR for predicting EV presence was greater than or equal to 3.5 with 61% sensitivity and 76% specificity.
Abu El Makarem et al. revealed that HOMA-IR score greater than or equal to 3 was optimal for the prediction of EVs with a resulting 88.6% sensitivity, 95.2% specificity, and 90% accuracy.
However, both Cammà et al., and Yousif et al. conducted studies only on Child A cirrhotic patients. Out of the 104 patients in Cammà et al. study, 63 patients have EV (84% was of grade I EV) and 20 of them had type 2 diabetes mellitus, by which the measurement of insulin sensitivity can be easily affected by medications for diabetes. Of the patients, 92% had genotype I HCV infection unlike our study in which genotype 4 mostly predominated. Although the study of Yousif and colleagues was conducted on Egyptian, nondiabetic patients, only 79% of patients have EVs and the majority of them were of grade II EVs.
Eslam et al. showed that HOMA-IR cutoff level greater than 4 had 98.7% sensitivity and 71.0% specificity. The differences between the studies' results may be attributed to the different ethnic groups of patients. The Eslam et al. study is a multicentered study conducted in two centers (in Spain and Egypt), on patients with different etiological factors for cirrhosis including alcohol, HCV, HBV, and others; only 76% of patients had EVs.
There was a high statistically significant difference (P < 0.001) between the studied groups regarding portal vein diameter with a cutoff value greater than 13 mm. This goes with Sudha et al., who stated that portal vein diameter (>13 mm) is an independent noninvasive predictor of the presence of esophageal varices in patients with cirrhosis. Chandail et al. observed that any patient with portal vein size greater than 13.75 mm showed EV on endoscopy with 79.4% sensitivity and 88.2% specificity.
There was a high statistically significant difference (P < 0.001) between the studied groups regarding platelet/spleen bipolar diameter ratio with a cutoff value less than or equal to 1414. Abu El Makarem et al. concluded that a cutoff value of 939.7 had 100% sensitivity, 86.3% specificity, and 96.5% accuracy. Esmat et al. concluded that a cutoff value of 1326.58 had 96.34% sensitivity, 83.33% specificity, and 94% accuracy. Yousif et al. showed that a cutoff of 750 had a sensitivity of 81%, specificity of 81%, and an accuracy of 81%.
The difference in values between the studies may be due to factors influencing the platelet count including infection, bleeding, drugs, and the cause of portal hypertension (being in this study HCV-related cirrhosis only), in addition to the differences between the sonographers of different studies.
There was a high statistically significant difference (P < 0.001) between the studied groups regarding mid-clavicular liver span/albumin ratio with a cutoff value greater than 3.51. This goes with Adel and George, who stated that the best cutoff value was at 3.5 where the sensitivity was 80% and specificity was 70% and Yousif et al. revealed that the cutoff value greater than 3.5 was adequate for varices prediction with 78.5% sensitivity, 57.1% specificity, and 74% accuracy.
In another Egyptian study by Esmat et al., the cutoff value of 4.422 gave 91.46% sensitivity and 77.78% specificity.
Differences between values in the studies may be attributed to the different degrees of decompensation as patients in our study were Child A, B, and C with the mean albumin concentration in the range of 2.1–5.3 g/dl, from different ethnic backgrounds, in addition to the differences between the sonographers of different studies.
On a comparison between HOMA-IR, platelet count/splenic bipolar diameter ratio, mid-clavicular liver span/albumin concentration ratio and portal vein diameter as four noninvasive predictors of EV in the present study, HOMA-IR score gave the highest accuracy (92.7%). The next was the mid-clavicular liver span/albumin ratio (86.3%) followed by portal vein diameter (86.1%). The least accurate was platelet count/splenic bipolar diameter ratio (79.1%).
This agrees with Yousif et al. who stated that HOMA-IR score gave the highest accuracy followed by the platelet count/splenic bipolar diameter ratio (81%) at a cutoff value of 750 (sensitivity 81% and specificity 81%). The least accurate was the right liver lobe diameter/albumin ratio (74%) at a cutoff value of 3.5 with a sensitivity of 78.5% and a specificity of 57.1%.
Rizk et al. showed that for the diagnosis of any EV, HOMA-IR score of greater than 3.49 had 62% sensitivity and 74% specificity compared with 85% sensitivity and 62% specificity for platelet count/spleen ratio of less than 790. While for the diagnosis of large EV a HOMA-IR score of greater than 3.7 had 74% sensitivity and 60% specificity compared with 89.9% sensitivity and 74.3% specificity for platelet count/spleen ratio of less than 770.4.
| Conclusion|| |
From this study, we conclude that IR measured by HOMA-IR is a noninvasive parameter that can provide accurate information pertinent to the determination of the presence and grade of EV in patients with HCV-induced cirrhosis regardless of their Child classification. It can help physicians to restrict the use of endoscopic screening only to patients presenting a high probability of EV. This is especially useful in clinical settings where resources are limited and endoscopic facilities are not present in all areas. Such is the case in Egypt, where there is a large number of patients who require screening for EV.
Financial support and sponsorship
Conflicts of interest
There are no conflicts of interest.
| References|| |
Sudha K, Sudarsi B, Siddeswari R, Manohar S. Correlation of portal vein size with esophageal varices severity in patients with cirrhosis of liver with portal hypertension. IJSRP 2015; 5
Eslam M, Ampuero J, Jover M, Abd-Elhalim H, Rincón D, Shatat M, et al
. Predicting portal hypertension and variceal bleeding using non-invasive measurements of metabolic variables. Ann Hepatol 2013; 2
Maruyama H, Kobayashi K, Kiyono S, Yokosuka O. Interrelationship between insulin resistance and portal haemodynamic abnormality in cirrhosis. Int J Med Sci 2017; 14
Ishikawa T, Shiratsuki S, Matsuda T, Iwamoto T, Takami T, Uchida K, et al
. Occlusion of portosystemic shunts improves hyperinsulinemia due to insulin resistance in cirrhotic patients with portal hypertension. J Gastroenterol 2014; 49
Erice E, Llop E, Berzigotti A, Abraldes JG, Conget I, Seijo S, et al
. Insulin resistance in patients with cirrhosis and portal hypertension. Am J Physiol Gastrointestinal Liver Physiol 2012; 302
Zardi E, Di Matteo F, Pacella C, Sanyal A. Invasive and non-invasive techniques for detecting portal hypertension and predicting variceal bleeding in cirrhosis: a review. Ann Med 2013; 46
Fartoux L, Poujol-Robert A, Guéchot J, Wendum D, Poupon R, Serfaty L. Insulin resistance is a cause of steatosis and fibrosis progression in chronic hepatitis C. Gut 2005; 54
Miyajima I, Kawaguchi T, Fukami A, Nagao Y, Adachi H, Sasaki S, et al
. Chronic HCV infection was associated with severe insulin resistance and mild atherosclerosis: a population-based study in an HCV hyperendemic area. J Gastroenterol 2013; 48
Yousif M, Elhady H, Wahba M, Esh A. Insulin resistance as a non-invasive parameter for prediction of esophageal varices in patients with hepatitis C virus-induced cirrhosis. Zagazig Univ Med J 2014; 20
Neto DW, Tredicci MT, Coelho F, Makdissi F, Herman P. Portal pressure decrease after esophagogastric devascularization and splenectomy in schistosomiasis: long-term varices behavior, rebleeding rate, and role of endoscopic treatment. Arquivos Gastroenterol 2018; 55
Cammà C, Petta S, Di Marco V, Bronte F, Ciminnisi S, Licata G, et al
. Insulin resistance is a risk factor for esophageal varices in hepatitis C virus cirrhosis. Hepatology 2009; 49
Abu El Makarem M, Shatat M, Shaker Y, Abdel Aleem A, El Sherif A, Moaty M, et al
. Platelet count/bipolar spleen diameter ratio for the prediction of esophageal varices: the special Egyptian situation: noninvasive prediction of esophageal varices. Hepatitis Monthly 2011; 11
Chandail VS, Kotwal SK, Koul S, Gupta R, Mahajan A. Non-invasive markers for prediction of varices in patients with portal hypertension. Int J Res Med Sci 2017; 5
Esmat S, Omran D. Study of noninvasive predictors of portal hypertension in liver cirrhotic Egyptian patients. J Am Sci 2011; 7
Adel AM, George SR. A novel non-invasive ratio for oesophageal varices prediction in HCV liver cirrhosis Egyptian patients. J Am Sci 2011; 7
Rizk M, Abd- Elmaksoud M, Elasyouty I, Khalil F, Agha A, Dabour A. Insulin resistance and platelet count/splenic bipolar diameter ratio for predicting esophageal varices in hepatitis C virus Cirrhosis. Banha University Website. 2015. Available from: http://www.bu.edu.eg/staff/mahmoudrizk12-publications/26064
. [Last accessed on 2015].
[Table 1], [Table 2], [Table 3], [Table 4]