SEARCH

SEARCH BY CITATION

Abstract

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Indirect methods to predict the presence of esophageal varices (EV) in patients with cirrhosis are not sensitive enough to be used as a surrogate for endoscopy. We tested the effectiveness of liver stiffness measurement (LSM) by transient elastography and the presence of insulin resistance (IR), a marker associated with fibrosis progression, in the noninvasive prediction of portal hypertension. One hundred four consecutive patients with newly diagnosed Child A hepatitis C virus (HCV) cirrhosis underwent upper gastrointestinal endoscopy to search for EV. Clinical, anthropometric, biochemical, ultrasonographic, and metabolic features, including IR by the homeostasis model assessment (HOMA), and LSM by transient elastography, were recorded at the time of endoscopy. EVs were detected in 63 of 104 patients (60%). In 10 patients (16%), the EVs were medium-large (≥F2). By multivariate analysis, the presence of EVs was independently associated with a low platelet count/spleen diameter ratio (OR, 0.998; 95% CI, 0.996-0.999) and a high HOMA-IR score (OR, 1.296; 95%CI, 1.018-1.649), not with LSM (OR, 1.009; 95%CI, 0.951-1.070). It is noteworthy that nine of ten patients with medium-large EVs had a platelet/spleen ratio of less than 792 or an HOMA-IR of greater than 3.5. The independent association between low platelet count/spleen diameter ratio (OR, 0.998; 95%CI, 0.996-1.000), high HOMA-IR score (OR, 1.373; 95%CI, 1.014-1.859) and presence of EV was confirmed in the subgroup of 77 nondiabetic subjects. Conclusions: In patients with Child A HCV cirrhosis, two simple, easy-to-get tests, namely the platelet/spleen ratio and insulin resistance measured by HOMA-IR, regardless of the presence of diabetes, significantly predict the presence of EV, outweighing the contribution given by transient elastography. (HEPATOLOGY 2009;49:195-203)

Portal hypertension (PH), defined by a hepatic venous pressure gradient (HVPG) greater than 6 mmHg,1 is a common complication of cirrhosis. The presence and the development of esophageal varices (EV) is a clinical manifestation of PH,2, 3 with a prevalence that can range from 40% to 80% in patients with cirrhosis. This prevalence increases progressively in relation to the severity of liver damage.4, 5 The presence of EV is also a clear indicator of a certain stage of cirrhosis.6 The development of EV in patients with cirrhosis occurs when the HVPG is greater than 10 mmHg,3, 7 with an incidence of approximately 5% per year and a yearly rate of progression to larger varices of 5% to 15%.4, 5, 8 The clinical relevance of EV is linked to the risk of bleeding that occurs when HVPG is greater than 12 mmHg,7–9 with a mortality rate that exceeds 20% within 6 weeks from the bleeding episode, despite aggressive treatment.10

The Baveno IV 2005 Consensus Workshop11 and the American Association for the Study of Liver Diseases 2007 single-topic symposium on portal hypertension12 recommended that endoscopic screening for esophageal and gastric varices should always be performed when a diagnosis of cirrhosis is made. Upper endoscopy should be repeated at 2-year to 3-year intervals in patients without varices, and at 1-year to 2-year intervals in those with small varices, to evaluate their development or progression.11, 12 These guidelines might not be ideal for clinical practice, exposing patients with different prevalence of EV to endoscopy-related complications, reported in approximately 0.1% of procedures.13 Furthermore, particularly in subjects with a lower prevalence of EV (those suffering from Child A cirrhosis), these recommendations imply a considerable burden of endoscopies and related costs. They also require that patients repeatedly undergo an unpleasant procedure, even though up to 50% of them may still not have developed EV 10 years after the diagnosis of cirrhosis.5

To reduce the number of endoscopies, many studies were carried out to identify features that may noninvasively predict the presence of EV of any size, or at least of medium-large size at higher risk of bleeding.14–28 These studies, performed in patients with compensated or decompensated cirrhosis, focused on clinical, biochemical, and instrumental (ultrasonography, computed tomography, endoscopic capsule) data, identifying single variables or combinations of variables able to predict EV. However, the accuracy of prediction was suboptimal in all settings, making it inadvisable to adopt a noninvasive screening policy for EV in clinical practice.29 Two recent studies have found linear relationships between portal hypertension, EV, and liver stiffness measurement (LSM) evaluated by transient elastography (TE),30, 31 thus suggesting TE as a complementary technique in the assessment of PH.

Insulin resistance (IR) is exceedingly common in patients with hepatitis C virus (HCV)-related chronic liver disease,32 and both experimental33 and clinical studies34 suggested that HCV per se is able to decrease insulin sensitivity. IR has been systematically associated with advanced fibrosis and fibrosis progression in several reports.32, 34–37 Although the underlying mechanisms linking hyperinsulinemia/IR to fibrosis are far from clear, the quantitative measurement of IR might be a potential predictor of portal hypertension in early cirrhosis, alone or in combination with other clinical features.

We report a prospective study in patients with Child A HCV cirrhosis aimed at evaluating whether LSM and metabolic factors, particularly IR, correlate with the presence of EV, in search of an accurate noninvasive model for predicting the presence of EV.

Patients and Methods

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Patients.

We enrolled in the study 104 newly diagnosed patients with Child A HCV cirrhosis, consecutively observed at the GI & Liver Unit at the University Hospital of Palermo, Italy, between January 2006 and March 2007, and fulfilling all criteria detailed below. Patients were included if they had a diagnosis of HCV cirrhosis based on liver biopsy or history, physical examination, and biochemical parameters. Exclusion criteria were: (1) advanced cirrhosis (Child-Pugh classes B and C); (2) other causes of liver disease or mixed causes (alcohol abuse, hepatitis B, autoimmune liver disease, Wilson's disease, hemochromatosis, α1-antitrypsin deficiency); (3) human immunodeficiency virus infection; (4) current or previous history of ascites or hepatic encephalopathy or portal hypertensive bleeding; (5) hepatocellular carcinoma; (6) portal thrombosis; (7) current treatment with any dosage of insulin; (8) previous or current treatment with beta-blockers, diuretics, or other vasoactive drugs; (9) parenteral drug addiction or alcohol abuse in the last year.

The study was performed in accordance with the principles of the Declaration of Helsinki and its appendices and with local and national laws. Approval was obtained from the hospital's Internal Review Board and Ethics Committee, and by written informed consent from all patients.

Clinical and Laboratory Assessment.

The following data were collected at the time of recruitment: age, sex, weight, and height. Body mass index (BMI) was calculated as weight in kilograms/height in square meters. Patients with a BMI of 18.5 to 24.9 kg/m2 were classified as normal, those with a BMI of 25 to 29.9 as overweight, those with a BMI of 30 or more as obese. The diagnosis of type 2 diabetes was based on the American Diabetes Association revised criteria, using a value of fasting blood glucose of 126 mg/dL or greater on at least two occasions, or ongoing treatment with hypoglycemic agents. In patients with a previous diagnosis of type 2 diabetes, current therapy with insulin or oral agents was documented.

The diagnosis of arterial hypertension was based on the following criteria: systolic blood pressure 140 mm Hg or higher or diastolic blood pressure 90 mm Hg or higher (measured three times within 30 minutes in the sitting position using a brachial sphygmomanometer) or ongoing antihypertensive pharmacological treatment after a diagnosis of arterial hypertension.

A 12-hour overnight fasting blood sample was drawn at the time of recruitment to determine the serum levels of alanine aminotransferase, gamma-glutamyltransferase, total cholesterol, high-density lipoprotein cholesterol, triglycerides, ferritin, plasma glucose concentration, and platelet count. Serum insulin was determined by a two-site enzyme enzyme-linked immunosorbent assay (Mercodia Insulin ELISA, Arnika). IR was determined by the homeostasis model assessment (HOMA) method by using the following equation38: Insulin resistance (HOMA-IR)= Fasting insulin (μU/mL) × fasting glucose (mmol/L)/22.5. HOMA-IR has been validated in comparison with the euglycemic/hyperinsulinemic clamp technique in both diabetic and nondiabetic subjects.39

HCV RNA was tested at the time of recruitment by qualitative polymerase chain reaction (Cobas Amplicor HCV Test version 2.0; limit of detection, 50 IU/mL) and quantified by Versant HCV RNA 3.0 bDNA (Bayer Co. Tarrytown, NY) expressed in IU/mL. Genotyping was performed by INNO-LiPA, HCV II, Bayer.

Instrumental Assessment.

After an overnight fast, all patients underwent an ultrasound examination to evaluate the signs of portal hypertension (portal vein flux and diameter, respiratory variations, and flux of splenic and mesenteric veins, splenomegaly, and ascites). Immediately after, TE was performed using a FibroScan apparatus (Echosens, Paris, France) to measure LSM. The operator was a staff physician (F.B.) who had previously performed at least 100 determinations in patients with chronic liver disease. The median value of 10 successful acquisitions, expressed in kilopascal (kPa), was kept as representative of LSM. As previously described in the literature,40 and as suggested by the manufacturing company, we considered 10 successful acquisitions with a success rate of at least 60% and with an interquartile range lower than 20% as representative measurements.

On the same day, upper gastrointestinal endoscopy was performed by two endoscopists (S.P. and F.S.) working together and unaware of clinical, laboratory, and ultrasonographic data. The size of the varices was determined according to The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices.41 Discrepancies between examiners were infrequent (the k value of the interobserver agreement for the size of EV was 0.80) and were resolved by discussion.

Statistical Methods.

Continuous variables were summarized as mean ± standard deviation and categorical variables as frequency and percentage. Multiple logistic regression models were used to assess the relationship of EV presence to demographic, clinical, biochemical, metabolic, and instrumental features of the patients in the entire population and in the subgroup of nondiabetic subjects. In these models, the dependent variable was EV coded as 0 = absent or 1 = present. As candidate risk factors for EV presence we selected age, sex, BMI, baseline aspartate aminotransferase, alanine aminotransferase, bilirubin, international normalized ratio, creatinine, platelet count, gamma-glutamyltransferase, ferritin, total cholesterol, high-density lipoprotein cholesterol, triglycerides, blood glucose and insulin levels, HOMA score, Child-Pugh score, portal vein and spleen diameter, mesenteric and splenic vein respiratory variations, platelet count/spleen diameter ratio, LSM, diabetes (evaluated only in the entire population), and arterial hypertension.

Variables found to be associated with the dependent variable at univariate logistic regression at a probability threshold of less than 0.10 were entered into multivariate logistic regression models. To avoid the effect of colinearity, HOMA-IR score, blood glucose levels, insulin levels, and diabetes, as well as platelet count, spleen diameter, and platelet count/spleen diameter ratio, were not included in the same multivariate model. Regression analyses were performed using PROC LOGISTIC and subroutine in SAS (SAS Institute, Inc., Cary, NC).42

Receiver operating characteristic (ROC) curves were applied to find the best cutoff values, and to identify the area under ROC curve (AUC) of the individual variables independently associated with the presence of EV. Significant variables in the multivariate analysis were also used to generate a prediction rule. For each case a score was calculated and a probability of response assigned, giving a set of values for the variables. In this model, the predictive role of each candidate predictor is evaluated by the following expression:

  • equation image

where P(Xkn) = the likelihood of event (in this case, the presence of EV) in the examined series of n patients characterized by the set of variables Xk; n = 1, 2,… 104.

α = log-odds of event likelihood for a patient with a standard set of variable (Xkn = 0).

Xkn = vector of variables X0n, X1n,… Xkn for the n-th patients. k 5 0, 1, 2.

Âk = vector of parameters 0, 1,… k that weighs the contribution of each variable to the likelihood of event. Σmath imageβkXkn = sum of the products of parameter k by the variables Xkn of the n-th patient. We performed an ROC curve, which shows the capacity of the related model to discriminate between patients with EV and those without EV.

Results

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

Patient Features.

Baseline features of all 104 patients are shown in Table 1. Sixty were males, mean age was 61 ± 9, and mean BMI was 26.0 ± 3.2 kg/m2. Overall, 45 of 104 patients (43.2%) had normal weight, 43 of 104 (41.3%) were overweight, and 11 of 114 (15.5%) were obese. The mean value for HOMA was high (3.98), and 27 patients (26%) were diabetic. Mean values for total and high-density lipoprotein cholesterol and triglycerides were within the normal range, even in diabetic patients.

Table 1. Demographic, Laboratory, Metabolic and Instrumental Features of 104 Consecutive Patients with Child A HCV Cirrhosis
VariablePatients with HCV Cirrhosis (n = 104)
  1. Abbreviations: yrs, years; IU, international units; INR, international normalized ratio; HDL, high-density lipoprotein; HOMA, homeostasis model assessment.

Mean age, years61.4 ± 9.5
Sex 
 Male60 (57.7)
 Female44 (42.3)
Mean Body Mass Index, kg/m226.0 ± 3.2
Body Mass Index, kg/m2 
 <2545 (43.2)
 25-29.943 (41.3)
 ≥3011 (15.5)
Aspartate aminotransferase, IU/L (NL < 30 IU/L)98.5 ± 64.5
Alanine aminotransferase, IU/L (NL < 30 IU/L)111.3 ± 73.3
Albumin, g/dL3.9 ± 0.3
Bilirubin, mg/dL0.98 ± 0.46
INR1.07 ± 0.07
Child-Pugh Score5.19 ± 0.39
Child-Pugh Score 
 A 584 (80.8)
 A 620 (19.2)
Creatinine, mg/dL0.79 ± 0.47
Portal vein (diameter), mm11.9 ± 1.8
Mesenteric and splenic vein respiratory variations 
 Present9 (8.7)
 Absent95 (91.3)
Spleen (longitudinal diameter), cm15.0 ± 3.2
Platelet count × 103/mm103.8 ± 50.2
Platelet count/spleen diameter ratio732.95 ± 392.84
Liver Stiffness, KPa24.4 ± 11.0
Esophageal varices 
 Present63 (60.6)
 Absent41 (39.4)
 Small53 (84.1)
 Medium/large10 (15.9)
Gamma glutamyl transferase, IU107.4 ± 100.7
Ferritin, ng/mL286.0 ± 299.2
Cholesterol, mg/dL144.3 ± 37.2
HDL cholesterol, mg/dL51.4 ± 15.3
Triglycerides, mg/dL103.4 ± 41.7
Blood glucose, mg/dL106.6 ± 36.1
Insulin, μU/mL15.32 ± 8.08
HOMA-score3.98 ± 2.75
Type 2 diabetes 
 Present27 (26.0)
 Absent77 (74.0)
Arterial hypertension 
 Present17 (16.3)
 Absent87 (83.7)
Viral genotype 
 196 (92.3)
 26 (5.7)
 31 (1)
 41 (1)

All patients were Child-Pugh's class A. Cirrhosis was defined histologically in only 15 patients (13.1%). Sixty percent of subjects (63/104) had EV; they were of a medium-large size (≥F2) in 15% of cases (10/63). Four patients had gastric varices; all of them also with EV. Ninety-two percent of patients (96/104) had genotype 1 HCV infection (Table 1).

Mean LSM value was 24.4 kPa (range, 6.6-59.3). Thirteen patients (12%) had an unsuccessful LSM. These cases were considered a failure of the method, the main cause being overweight or obesity (all had a BMI of >28).

Factors Associated with EV.

Univariate and multivariate comparisons of variables between patients with and those without EV are reported in Table 2. The absence of respiratory variations of the mesenteric and splenic veins, high spleen diameter, low platelet counts, and platelet count/spleen diameter ratio, high insulin levels, and HOMA-IR score, and high LSM were all associated with the presence of EV (P < 0.10). However, multivariate logistic regression analysis showed that only two features were independently linked to the presence of EV: high HOMA-IR [odds ratio (OR), 1.296; 95% confidence interval (95%CI), 1.018-1.649; P = 0.03] and low platelet count/spleen diameter ratio (OR, 0.998; 95%CI, 0.996-0.999; P = 0.001).

Table 2. Univariate and Multivariate Analysis of Factors Associated with Presence of Esophageal Varices in 104 Consecutive Patients with Child A HCV Cirrhosis
VariablePatients Without Esophageal Varices N = 41Patients with Esophageal Varices N = 63Univariate Analysis P ValueMultivariate Analysis OR (95% CI) P Value
  1. Abbreviations: yrs, years; IU, international units; INR, international normalized ratio; HDL, high density lipoprotein; HOMA, homeostasis model assessment.

Age, years62.9 ± 9.660.4 ± 9.40.18
Sex    
 Male21(51.2)39 (61.9)  
 Female20 (48.8)24 (38.1)0.28
Body Mass Index, kg/m225.8 ± 2.426.1 ± 3.60.74
Aspartate aminotransferase, IU/L100.4 ± 74.297.3 ± 58.10.80
Alanine aminotransferase, IU/L108.4 ± 75.7113.2 ± 72.20.74
Albumin, g/dL4.02 ± 0.433.95± 0.310.34
Bilirubin, mg/dL0.94 ± 0.481.00± 0.450.52
INR1.06 ± 0.071.08± 0.070.11
Child-Pugh Score5.2 ± 0.45.1± 0.30.57
Creatinine, mg/dL0.77 ± 0.380.80± 0.520.74
Portal vein (diameter), mm11.6 ± 1.712.0 ± 1.80.27
Mesenteric and splenic vein respiratory variations    
 Present7 (20.6)2 (3.2)  
 Absent34 (79.4)61 (96.8)0.020.385 (0.059-2.491)0.31
Spleen (longitudinal diameter), cm13.8 ± 3.415.7 ± 2.90.005
Platelet count n/mm125.8 ± 54.989.2 ± 41.20.001
Platelet count/spleen diameter ratio932.7 ± 408.3601.9 ± 323.60.00020.998 (0.996-0.999)0.001
Liver stiffness, KPa20.4 ± 10.027.6 ± 10.90.0011.009 (0.951-1.070)0.77
Gamma glutamyl transferase, IU113.9 ± 130.6103.1 ± 76.10.59
Ferritin, ng/mL334.3 ± 356.5254.6 ± 253.90.22
Cholesterol, mg/dL144.7 ± 38.8144.0 ± 36.50.93
HDL cholesterol, mg/dL51.7 ± 16.351.2 ± 14.80.87
Triglycerides, mg/dL95.3 ± 37.7108.6 ± 43.60.11
Blood glucose, mg/dL100.1 ± 32.0110.9 ± 38.20.14
Insulin, μU/mL13.01 ± 6.6816.82 ± 8.590.02
HOMA score3.15 ± 2.174.50 ± 2.960.011.296 (1.018-1.649)0.03
Type 2 diabetes    
 Present7 (17.0)20 (31.7)  
 Absent34 (83.0)43 (68.3)0.10
Arterial hypertension    
 Present5 (12.2)12 (19.0)  
 Absent36 (87.8)51 (81.0)0.35

A multivariate analysis (Table 3) performed on the subset of nondiabetic patients again confirmed high HOMA-IR (OR: 1.373; 95% CI; 1.014-1.859; P = 0.04) and low platelet count/spleen diameter ratio (OR: 0.998; 95% CI; 0.996-1.000; P = 0.01) as significant predictors of EV presence. Comparable results were observed restricting the analysis to the subset of genotype 1 HCV patients.

Table 3. Multivariate Analysis of Factors Associated with Presence of Esophageal Varices in 77 Consecutive Nondiabetic Patients with Child A HCV Cirrhosis
VariableMultivariate Analysis
OR (95% CI)P Value
  1. yrs, years; HOMA, homeostasis model assessment.

Sex, male/female1.087 (0.310-3.820)0.89
Age, years0.957 (0.902-1.014)0.13
Mesenteric and splenic vein respiratory variations0.448 (0.059-3.397)0.43
Platelet count/spleen diameter ratio0.998 (0.996-1.000)0.01
HOMA score1.373 (1.014-1.859)0.04
Liver stiffness, Kpa1.008 (0.945-1.076)0.80

ROC curves analysis (Fig. 1) identified a platelet count/spleen ratio of greater than 792 [AUC, 0.740; standard error (SE), 0.054; 95%CI, 0.639-0.825; sensitivity, 83%; specificity, 60%; positive likelihood ratio, 2.08; negative likelihood ratio, 0.28] and an HOMA-IR score of greater than 3.5 (AUC, 0.671; SE, 0.055; 95%CI, 0.567-0.765; sensitivity, 61%; specificity, 76%; positive likelihood ratio, 2.58; negative likelihood ratio, 0.51) as the best cutoff for predicting the presence of EV (Fig. 2). The discriminating ability of the rule generated by the model was high (AUC, 0.800; SE, 0.046; 95%CI, 0.695-0.871; sensitivity, 75%; specificity, 75%; positive likelihood ratio, 2.98; negative likelihood ratio, 0.34). The prevalence of EV gradually decreased from the “worst” to the “best” class for each variable (Fig. 3). It is noteworthy that only 3 of 41 patients (7.3%) without EV were in the worst class (false positive), and that 4 of 63 patients (6.3%) with EV were in the best class (false negative). Particularly among the 10 patients with medium-large EV only one subject with diabetes (10%) had a platelet count/spleen ratio of greater than 792 and an HOMA score of less than 3.5.

thumbnail image

Figure 1. Receiver operating characteristic (ROC) curve for platelet count/spleen diameter ratio (dashed line) (AUC, 0.740; SE, 0.054; 95%CI, 0.639-0.825) and for HOMA score (solid line) (AUC, 0.671; SE, 0.055; 95%CI, 0.567-0.765) on the basis of the presence of any esophageal varices.

Download figure to PowerPoint

thumbnail image

Figure 2. Evaluation of the rule predicting the presence of esophageal varices by receiver operating features (ROC) curve (AUC, 0.800; SE, 0.046; 95%CI, 0.695-0.871).

Download figure to PowerPoint

thumbnail image

Figure 3. Prevalence of esophageal varices according to specific patterns of predictors. In parentheses the number of patients in each class. 1 = HOMA of >3.5, platelet count/spleen diameter ratio of <792; 2 = HOMA of <3.5, platelet count/spleen diameter ratio of <792; 3 = HOMA of >3.5, platelet count/spleen diameter ratio of >792; 4 = HOMA of <3.5, platelet count/spleen diameter ratio of >792.

Download figure to PowerPoint

Cost Analysis of Endoscopic Screening for Detection of Esophageal Varices.

On the basis of our prediction rule, our prescreening strategy was 20% more cost effective than those put forth in the current guidelines (Table 4).

Table 4. Cost Analysis of Endoscopic Screening for Detection of Esophageal Varices (EV) in 104 Child A HCV-Related Patients with Cirrhosis
StrategyNumber of Patients Who Underwent Endoscopy (%)Cost of Strategy ($)*Number of Patients with EV IdentifiedCost of EV Detection per Patient ($)
  • *

    Cost of the endoscopies performed for each strategy. Direct cost of diagnostic endoscopy was $800.

  • Ratio of the cost of strategy per number of patients with EV identified.

Endoscopy-for-all strategy104 (100)83,200631320
Prediction rule strategy84 (80.7)67,200591138
Ideal strategy63 (60)50,40063800

Discussion

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

In our setting we identified the platelet count/spleen diameter ratio, a well-known marker of EV, and the HOMA-IR score, an easy biochemical marker, as the only factors independently predicting the presence of EV, whereas LSM did not contribute significantly to risk assessment.

The Baveno IV 2005 Consensus Workshop11 and the American Association for the Study of Liver Diseases 2007 single-topic symposium on portal hypertension12 have recommended endoscopic screening for EV in patients with cirrhosis, regardless of Child class and cause. This policy is expensive, and it would be useful to have noninvasive predictors of EV in compensated Child A patients, at low risk of portal hypertension, as a pre-endoscopy screening tool. This goal was reasonably attained in the current study.

Several studies have shown that ascites,20 presence of spider nevi,24 low prothrombin time,18 alanine aminotransferase, and albumin levels,24 high portal vein diameter,18 splenomegaly,14, 17, 20, 22 and low platelet count14–18, 20–23 could serve as predictors of EV presence. However, all of these studies were quite heterogeneous, enrolling patients with cirrhosis of different causes (viral, alcoholic, and mixed) and different disease severity (Child B or end-stage liver disease). In our population of HCV patients with Child A cirrhosis, without biochemical and clinical alterations caused by poor liver function, we found no association between classic liver tests and the presence of EV. However, we confirmed the independent association of a lower platelet count/spleen ratio with the presence of EV, previously reported by Giannini et al.19 in a cohort of subjects at any stage of cirrhosis due to HCV infection or alcohol abuse. The platelet count/spleen ratio cutoff value with the best sensitivity and specificity for the diagnosis of EV identified by Giannini et al. was 909. Conversely, in our cohort we identified a value of 792 as the best cutoff. These different results are perhaps related to differences in etiology and class of disease between the two populations.

A novel finding of our work, not specifically evaluated in other studies, is the association of IR, regardless of diabetes, with the presence of EV. Our population had a low prevalence of obesity (15%), but nonetheless 26% of patients had diabetes, and the mean HOMA-IR score was approximately 4, higher than observed in our previous studies of subjects with chronic hepatitis C.32, 36 The high prevalence of diabetes and the high HOMA-IR values were not likely attributable to a “hepatogenous diabetes,”43 considering that they were found in a cohort of Child A patients. We identified an HOMA-IR score of greater than 3.5 as the cutoff value with the best sensitivity and specificity for predicting EV presence.

Several studies in chronic liver diseases have shown a strong and independent pathogenic link between IR and HCV infection34, 37 and between IR and the severity of hepatic fibrosis.32, 34, 37 Although our work was not designed to clarify the pathogenic interaction between IR and presence of EV, a few hypotheses can be put forward. Insulin is able to modulate the endothelial synthesis of nitric oxide and endothelin,44, 45 to induce the production of tumor necrosis factor alpha and connective growth factor, and to stimulate hepatic stellate cells,46, 47 the effectors in the pathogenesis of liver fibrosis and PH.48 Therefore, insulin could contribute to the pathogenesis of PH by interfering with both mechanical and dynamic mechanisms leading to collagen deposition, vasoconstriction, and regulation of sinusoidal structure.

In our study, LSM values were directly related to the presence of EV, confirming two recent studies that found a correlation between LSM and the presence and grade of EV,30 and between LSM and the presence of both EV and PH (evaluated by the hepatic vein pressure gradient).31 However, only univariate comparison between EV, PH, and LSM were carried out in these, and other factors were not considered. We were unable to demonstrate an association between LSM and the presence of EV by multivariate analysis. The relatively small number of patients included in our cohort, and differences among studies in terms of demographic features, baseline severity, and cause of disease may explain these conflicting results, as well as differences in the statistical methods. On the practical side, however, it is remarkable that two simple, easy-to-get tests are able to outperform TE, a costly technology still rarely available at most hospitals and not universally standardized and validated in this setting.

The discrimination ability of our prediction rule was very high, and the AUC of the model was 0.80. Prognostic models, although valuable in predicting the average probability of specific clinical events occurring in a group of patients, are much less accurate in predicting the outcome probability in individual patients. Therefore, care should be taken in transferring the good discriminating ability of our model into a widespread clinical application, unless validated in an independent setting.

In our study, the false-negative rate (the proportion of patients with varices who would be erroneously excluded from endoscopy because they had both platelet count/spleen ratio and HOMA-IR under the cutoff) was 6.3%, and the false-positive rate (the proportion of patients without varices who would erroneously undergo endoscopy after model prediction) was 7.3%. Furthermore, in patients in the best class (i.e., subjects with both platelet count/spleen ratio and HOMA under the cutoff), the likelihood of EV was only 20% (4/20). However, three of these four subjects had small varices without red signs, whereas only one had medium-sized varices. This patient was diabetic, and there is some concern on the use of HOMA-IR in the presence of long-standing diabetes. However, a diagnosis of diabetes is per se expression of IR, and could be considered a risk factor for EV. From a practical point of view, on the basis of our analysis we suggest that a platelet count/spleen ratio greater than 792 and an HOMA score less than 3.5 (if nondiabetic) could be useful to identify patients at low risk of EV. However, only after external validation in large, independent settings, these data might be used to avoid endoscopic screening.

In terms of endoscopy-sparing ability, our prescreening strategy was 20% more cost effective than those put forth in the current guidelines. Recently Spiegel et al.,49 on the basis of a Markovian modeling decision analysis, suggested that empiric beta-blocker therapy for the primary prophylaxis of variceal hemorrhage is a more cost-effective measure than endoscopic screening. However, this model considered Child A and B patients with cirrhosis of any etiology. Thus, in the setting of Child A HCV cirrhosis, the introduction of a noninvasive prediction rule could help in further refining this strategy.

The study has limitations. First, the analysis was carried out in a relatively small number of patients, and it will be interesting to determine whether this association holds true also in larger groups of patients with HCV cirrhosis and in patients with liver disease of other origins. Second, the cohort of Child A HCV patients with cirrhosis, at low prevalence of obesity, was enrolled in a tertiary referral center for liver disease, limiting the broad application of the results. A further methodological issue resides in the inability to dissect the temporal relation between IR and varices. Another limitation lies in the fact that cirrhosis was diagnosed on clinical grounds in most cases, without the confirmation of a liver biopsy. Lack of data on other variables, such as direct measurement of portal hypertension by HVPG, also could affect the interpretation of our findings. Finally, we cannot exclude the possibility that hidden abuse of alcohol may be responsible for the presence of EV in a few subjects.

In conclusion, our data confirm the independent association between low platelet count/spleen ratio and the presence of EV in patients with Child A HCV cirrhosis and identify a high HOMA-IR score as a new independent predictor of the presence of EV, outweighing the contribution of LSM. On the practical side, it is remarkable that two simple, easy-to-get tests might be able to improve the efficacy of endoscopic screening of EV, provided they are confirmed in independent settings.

Acknowledgements

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References

The authors thank Warren Blumberg for his help in editing this paper.

References

  1. Top of page
  2. Abstract
  3. Patients and Methods
  4. Results
  5. Discussion
  6. Acknowledgements
  7. References
  • 1
    Grace ND, Groszmann RJ, Garcia-Tsao G, Burroughs AK, Pagliaro L, Makuch RW, et al. Portal hypertension and variceal bleeding: an AASLD single topic symposium. HEPATOLOGY 1998; 28: 868880.
  • 2
    De Franchis R, Primignani M. Natural history of portal hypertension in patients with cirrhosis. Clin Liver Dis 2001; 5: 645663.
  • 3
    Groszmann RJ, Garcia-Tsao G, Bosch J, Grace ND, Burroughs AK, Planas R, et al. Beta-blockers to prevent gastroesophageal varices in patients with cirrhosis. N Engl J Med 2005; 353: 22542261.
  • 4
    Merli M, Nicolini G, Angeloni S, Rinaldi V, De Santis A, Merkel C, et al. Incidence and natural history of small esophageal varices in cirrhotic patients. J Hepatol 2003; 38: 266272.
  • 5
    D'Amico G, Pasta L, Madonia S, Tarantino G, Mancuso A, Malizia G, et al. The incidence of esophageal varices in cirrhosis. Gastroenterology 2001; 120: A2
  • 6
    D'Amico G, Garcia-Tsao G, Pagliaro L. Natural history and prognostic indicators of survival in cirrhosis: a systematic review of 118 studies. J Hepatol 2006; 44: 217231.
  • 7
    Garcia-Tsao G, Groszmann RJ, Fisher RL, Conn HO, Atterbury CE, Glickman M. Portal pressure, presence of gastroesophageal varices and variceal bleeding. HEPATOLOGY 1985; 5: 419424.
  • 8
    de Franchis R. Evaluation and follow-up of patients with cirrhosis and oesophageal varices. J Hepatol 2003; 3: 361363.
  • 9
    Perellò A, Escorsell A, Bru C, Gilabert R, Moitinho E, Garcia-Pagàn JC, et al. Wedged hepatic venous pressure adequately reflects portal pressure in hepatitis C virus-related cirrhosis. HEPATOLOGY 1999; 30: 13931397.
  • 10
    D'Amico G, De Franchis R, and a cooperative study group. Upper digestive bleeding in cirrhosis: post-therapeutic outcome and prognostic indicators. HEPATOLOGY 2003; 38: 599612.
  • 11
    De Franchis R. Evolving consensus in portal hypertension. Report of the Baveno IV consensus workshop on methodology of diagnosis and therapy in portal hypertension. J Hepatol 2005; 43: 167176.
  • 12
    Garcia-Tsao G, Sanyal AJ, Grace ND, Carey W, and the Pratice Guidelines Committee of the American Association for the Study of Liver Diseases, the Practice Parameters Committee of the American College of Gastroenterology. Prevention and management of gastroesophageal varices and variceal hemorrhage in cirrhosis. HEPATOLOGY 2007; 46: 922938.
  • 13
    Eisen GM, Baron TH, Dominitz JA, Faigel DO, Goldstein JL, Johanson JF, et al. American Society for Gastrointestinal Endoscopy. Complications of upper GI endoscopy. Gastrointest Endosc 2002; 55: 784793.
  • 14
    Chalasani N, Imperiale TF, Ismail A, Sood G, Carey M, Mel Wilcox C. Predictors of large esophageal varices in patients with cirrhosis. Am J Gastroenterol 1999; 94: 32853291.
    Direct Link:
  • 15
    Pilette C, Oberti F, Aub C, Rousselet MC, Bedossa P, Gallois Y, et al. Non-invasive diagnosis of esophageal varices in chronic liver diseases J Hepatol 1999; 31: 867873.
  • 16
    Zaman A, Hapke R, Flora K, Rosen HR, Benner K. Factors predicting the presence of esophageal or gastric varices in patients with advanced liver disease. Am J Gastroenterol 1999; 94: 32923296.
    Direct Link:
  • 17
    Madhotra R, Mulcahy HE, Willner I, Reuben A. Prediction of esophageal varices in patients with cirrhosis. J Clin Gastroenterol 2002; 34: 8185.
  • 18
    Schepis F, Camma C, Niceforo D, Magnano A, Pallio S, Cinquegrani M, et al. Which patients with cirrhosis should undergo endoscopic screening for esophageal varices detection? HEPATOLOGY 2001; 33: 333338.
  • 19
    Giannini E, Botta F, Borro P, Risso D, Romagnoli P, Fasoli A. Platelet count/spleen diameter ratio: proposal and validation of a non-invasive parameter to predict the presence of oesophageal varices in patients with liver cirrhosis. Gut 2003; 52: 12001205.
  • 20
    Thomopoulos KC, Labropoulou-Karatza C, Mimidis KP, Katsakoulis EC, Iconomou G, Nikolopoulou VN. Non-invasive predictors of the presence of large oesophageal varices in patients with cirrhosis. Dig Liver Dis 2003; 35: 473478.
  • 21
    Zein CO, Lindor KD, Angulo P. Prevalence and predictors of esophageal varices in patients with primary sclerosing cholangitis. HEPATOLOGY 2004; 39: 203209.
  • 22
    Sharma SK, Aggarwal R. Prediction of large esophageal varices in patients with cirrhosis of the liver using clinical, laboratory and imaging parameters. J Gastroenterol Hepatol 2007; 22: 19091915.
  • 23
    Levy C, Zein CO, Gomez J, Soldevila-Pico C, Firpi R, Morelli G, et al. Prevalence and predictors of esophageal varices in patients with primary biliary cirrhosis. Clin Gastroenterol Hepatol 2007; 5: 803808.
  • 24
    Berzigotti A, Gilabert R, Abraldes JG, Nicolau C, Bru C, Bosch J, et al. Noninvasive prediction of clinically significant portal hypertension and esophageal varices in patients with compensated liver cirrhosis. Am J Gastroenterol 2008; 103: 11591167.
    Direct Link:
  • 25
    Perri R, Chiorean MV, Fidler JL, Fletcher JG, Talwalkar JA, Stadheim L, et al. A prospective evaluation of computerized tomographic scanning as a screening modality for esophageal varices. HEPATOLOGY 2008; 47: 15871594.
  • 26
    de Franchis R, Eisen GM, Laine L, Fernandez-Urien I, Herrerias JM, Brown RD, et al. Esophageal capsule endoscopy for screening and surveillance of esophageal varices in patients with portal hypertension. HEPATOLOGY 2008; 47: 15951603.
  • 27
    Eisen GM, Eliakim R, Zaman A, Schwartz J, Faigel D, Rondonotti E, et al. The accuracy of PillCam ESO capsule endoscopy versus conventional upper endoscopy for the diagnosis of esophageal varices: a prospective three-center pilot study. Endoscopy 2006; 38: 3135.
  • 28
    Thabut D, Trabut JB, Massard J, Rudler M, Muntenau M, Messous D, et al. Non-invasive diagnosis of large oesophageal varices with FibroTest in patients with cirrhosis: a preliminary retrospective study. Liver Int 2006; 26: 271278.
  • 29
    D'Amico G, Morabito A. Noninvasive markers of esophageal varices: another round, not the last. HEPATOLOGY 2004; 39: 3034.
  • 30
    Kazemi F, Kettaneh A, N'Kontchou G, Pinto E, Ganne-Carrie N, Trinchet JC, et al. Liver stiffness measurement selects patients with cirrhosis at risk of bearing large oesophageal varices. J Hepatol 2006; 45: 230235.
  • 31
    Vizzutti F, Arena U, Romanelli RG, Rega L, Foschi M, Colagrande S, et al. Liver stiffness measurement predicts severe portal hypertension in patients with HCV-related cirrhosis. HEPATOLOGY 2007; 45: 12901297.
  • 32
    Petta S, Cammà C, Di Marco V, Alessi N, Cabibi D, Caldarella R, et al. Insulin resistance, and diabetes, increase fibrosis in the liver of patients with HCV genotype 1 infection. Am Journal Gastroenterol 2008; 103: 11361144.
    Direct Link:
  • 33
    Shintani Y, Fujie H, Miyoshi H, Tsutsumi T, Tsukamoto K, Kimura S, et al. Hepatitis C virus infection and diabetes: direct involvement of the virus in the development of insulin resistance. Gastroenterology 2004; 126: 840848.
  • 34
    Moucari R, Asselah T, Cazals-Hatem D, Voitot H, Boyer N, Ripault MP, et al. Insulin resistance in chronic hepatitis C: association with genotypes 1 and 4, serum HCV RNA level, and liver fibrosis. Gastroenterology 2008; 134: 416423.
  • 35
    Camma C, Bruno S, Di Marco V, Di Bona D, Rumi M, Vinci M, et al. Insulin resistance is associated with steatosis in nondiabetic patients with genotype 1 chronic hepatitis C. HEPATOLOGY 2006; 43: 6471.
  • 36
    Petta S, Cammà C, Di Marco V, Alessi N, Barbaria F, Cabibi D, et al. Retinol-binding protein 4: A new marker of virus-induced steatosis in patients infected with HCV genotype 1. HEPATOLOGY 2008 sub epub.
  • 37
    Hui JM, Sud A, Farrell GC, Bandara P, Byth K, Kench JG, et al. Insulin resistance is associated with chronic hepatitis C virus infection and fibrosis progression. Gastroenterology 2003; 125: 16951704.
  • 38
    Matthews DR, Hosker JP, Rudenski AS, Naylor BA, Treacher DF, Turner RC. Homeostasis model assessment: insulin resistance and beta-cell function from fasting plasma glucose and insulin concentrations in man. Diabetologia 1985; 28: 412419.
  • 39
    Ikeda Y, Suehiro T, Nakamura T, Kumon Y, Hashimoto K. Clinical significance of the insulin resistance index as assessed by homeostasis model assessment. Endocr J 2001; 48: 8186.
  • 40
    Castera L, Vergniol J, Foucher J, Le Bail B, Chanteloup E, Haaser M, et al. Prospective comparison of transient elastography: Fibrotest, APRI and liver biopsy for the assessment of liver fibrosis in chronic hepatitis C. Gastroenterology 2005; 128: 343350.
  • 41
    The North Italian Endoscopic Club for the Study and Treatment of Esophageal Varices. Prediction of the first variceal hemorrhage in patients with cirrhosis of the liver and esophageal varices: a prospective multicenter study. N Engl J Med 1988; 319: 983989.
  • 42
    SAS Technical Report, SAS/STAS software: changes & enhancement, release 6.07. Cary, NC: SAS Institute Inc, 1992.
  • 43
    Blendis L, Brill S, Oren R. Hepatogenous diabetes: reduced insulin sensitivity and increased awareness. Gastroenterology 2000; 119: 18001802.
  • 44
    Vincent MA, Montagnani M, Quon MJ. Molecular and physiologic actions of insulin related to production of nitric oxide in vascular endothelium. Curr Diab Rep 2003; 3: 279288.
  • 45
    Iwakiri Y, Groszmann RJ. Vascular endothelial dysfunction in cirrhosis. J Hepatol 2007; 46: 927934.
  • 46
    Svegliati-Baroni G, Ridolfi F, Di Sario A, Casini A, Marucci L, Gaggiotti G, et al. Insulin and insulin-like growth factor-1 stimulate proliferation and type I collagen accumulation by human hepatic stellate cells: differential effects on signal transduction pathways. HEPATOLOGY 1999; 29: 17431751.
  • 47
    Paradis V, Perlemuter G, Bonvoust F, Dargere D, Parfait B, Vidaud M, et al. High glucose and hyperinsulinemia stimulate connective tissue growth factor expression: a potential mechanism involved in progression to fibrosis in nonalcoholic steatohepatitis. HEPATOLOGY 2001; 34: 738744.
  • 48
    Rockey DC. Hepatic fibrosis, stellate cells, and portal hypertension. Clin Liver Dis 2006; 10: 459479.
  • 49
    Spiegel BMR, Targownik L, Dulai GS, Karsan HA, Gralnek IM. Endoscopic screening for esophageal varices in cirrhosis: is it ever cost effective? HEPATOLOGY 2003; 37: 366377.