Patent Application
20190148004
The present invention relates to the assessment of the presence and/or severity of varices, including esophageal varices and gastric varices, in particular to the detection of large esophageal varices. More specifically, the present invention relates to a non-invasive method comprising measuring blood markers and/or obtaining physical data and optionally recovering data from an endoscopic capsule for assessing the presence and/or severity of esophageal or gastric varices.
The majority of patients who succumb to fibrosis or cirrhosis die due to complications of increased portal venous pressure, including variceal hemorrhage, ascites, hepatic encephalopathy, and the like. Indeed, severe fibrosis, especially cirrhosis, induces portal hypertension which, above a portal pressure level of 10 mmHg, provokes esophageal varices. Bleeding from ruptured esophageal varices is a major cause of mortality and economic burden in cirrhosis.
Primary prevention of first bleeding in large esophageal varices significantly reduces mortality. Therefore, the recommended work-up of cirrhotic patients includes systematic screening of large esophageal varices.
The gold standard method to diagnose large esophageal varices is upper gastro-intestinal endoscopy (UGIE). However, UGIE is somewhat limited by some constraints and notably the poor acceptance by the patients, due to the invasiveness of this method.
There is thus a need for non-invasive methods for diagnosing large esophageal varices.
Non-invasive diagnosis of large esophageal varices is currently not accurate enough to be adopted in practice. In particular, esophageal capsule endoscopy (ECE) presents a clinically significant probability of missed esophageal varices (i.e. false negative results) or of false positive results.
There is thus a need for a non-invasive method for diagnosing esophageal varices, which is more accurate than esophageal capsule endoscopy, and in particular which allows reducing missed esophageal varices.
Non-invasive diagnosis of liver fibrosis has gained considerable attention over the last 10 years as an alternative to liver biopsy. The first generation of simple blood fibrosis tests combined common indirect blood markers into a simple ratio, like APRI (Wai et al., Hepatology 2003) or FIB-4 (Sterling et al., Hepatology 2006). The second generation of calculated tests combine indirect and/or direct fibrosis markers by logistic regression, leading to a score, like Fibrotest (Imbert-Bismut et al., Lancet 2001), ELF score (Rosenberg et al., Gastroenterology 2004), FibroMeter™ (Cales et al., Hepatology 2005), Fibrospect™ (Patel et al., J Hepatology 2004), and Hepascore (Adams et al., Clin Chem 2005). For example, WO2005/116901 describes a non-invasive method for assessing the presence of a liver disease and its severity, by measuring levels of specific variables, including biological variables and clinical variables, and combining said variables into mathematical functions, generally binary mathematical function to provide a score result, often called “score of fibrosis”.
There is currently a need for a non-invasive diagnostic test for directly assessing the presence and/or severity of varices.
WO2014/190170 describes a non-invasive test for assessing hepatic vein pressure gradient (HVPG) in cirrhotic patients, and suggests that this test may be used for assessing the absence of varices. However, the non-invasive test of WO2014/190170 presents the drawback of using blood markers without clinical potential, because these markers, while commonly used for research purpose, may not easily be used for clinical diagnosis, due either to a difficult implementation or to the cost of the measurement. Moreover, there is no experimental demonstration in WO2014/190170 that the described non-invasive test may efficiently be used for assessing the presence of esophageal varices. Furthermore, the non-invasive test of WO2014/190170 only results in two situations: either the patient shows a HVPG lower than 12 mmHg, and is diagnosed as not presenting esophageal varices, either the patient shows a HVPG of at least 12 mmHg, and an additional test is required for assessing the presence of esophageal varices (usually endoscopy).
The non-invasive test of WO2014/190170 was developed on cirrhotic patients, i.e. in patients already diagnosed with cirrhosis. However, the construction and performance evaluation of non-invasive tests of cirrhosis are limited by the characteristics of liver biopsy which is an imperfect gold standard. Therefore, a non-invasive test for assessing the presence of esophageal varices should ideally circumvent the intermediate step of cirrhosis diagnosis.
There is thus a need for a non-invasive method for diagnosing esophageal varices without the drawbacks of the non-invasive tests of the prior art.
In the present invention, the Applicants develop a non-invasive method for diagnosing esophageal varices in a patient with a liver disease (whether or not this patient was previously diagnosed as cirrhotic), wherein said method comprises performing a non-invasive diagnostic test for assessing the severity of a hepatic condition, using cut-offs for assessing the presence of varices instead of cut-offs for assessing the severity of a hepatic condition. In one embodiment, the method of the invention further comprises combining in a score blood markers, clinical markers, and data obtained by esophageal capsule endoscopy.
Experimental data obtained by the Applicant demonstrate that the non-invasive method of the invention may be used in any liver disease patient, and strongly reduces the number of missed esophageal varices as compared to ECE for example.
The present invention thus relates to a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices in a liver disease patient, wherein said method comprises:
In one embodiment, the present invention relates to a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices in a liver disease patient, wherein said method comprises:
The present invention also relates to a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices in a liver disease patient, wherein said method comprises:
In one embodiment, the preset invention relates to a non-invasive method, wherein step a) comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and carrying out another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan), ARFI, VTE, supersonic elastometry and MRI stiffness, and optionally measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, the method is for assessing the presence of large esophageal varices.
In one embodiment, said cut-offs are a negative predictive value (NPV) cut-off and a positive predictive value (PPV) cut-off, or a sensitivity cut-off and a specificity cut-off.
In one embodiment, said NPV and PPV cut-offs define two predictive zones, a NPV predictive zone and a PPV predictive zone.
In one embodiment,
In one embodiment,
In one embodiment, if the value obtained in step (a) is in the indeterminate zone between the NPV cut-off and the PPV cut-off or between the sensitivity cut-off and the specificity cut-off, then the method further comprises one or more repetition of step (a) and step (b) wherein at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out, thereby defining new NPV and PPV predictive zones and assessing the presence and/or severity of varices in said patient through the use of multiple NPV and PPV predictive zones.
In one embodiment, if the value obtained in step (a) is in the indeterminate zone between the NPV cut-off and the PPV cut-off or between the sensitivity cut-off and the specificity cut-off, then the method further comprises the steps of:
In one embodiment, the imaging data on varices status are obtained by a non-invasive imaging method, preferably esophageal capsule endoscopy; or by a radiologic method, preferably a scanner.
In one embodiment, the non-invasive test carried out in step (a) is a blood test, preferably selected from ELF, FibroSpect™, APRI, FIB-4, Hepascore, Fibrotest™, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™; or a physical method, preferably selected from VCTE, ARFI, VTE, supersonic elastometry or MRI stiffness.
In one embodiment, at step (c), the obtained variables are the variables of the non-invasive test carried out in step (a).
In one embodiment, the non-invasive test carried out in step (a) is a CirrhoMeter.
In one embodiment, the non-invasive method of the invention comprises carrying out at least two non-invasive tests for assessing the severity of a hepatic lesion or disorder, wherein said at least two non-invasive tests are different.
In one embodiment, the non-invasive test carried out in step (a) is a CirrhoMeter, and wherein the variables obtained at step (c) are the variables of a CirrhoMeter.
In one embodiment, the patient is affected with a chronic hepatic disease, preferably selected from the group comprising chronic viral hepatitis C, chronic viral hepatitis B, chronic viral hepatitis D, chronic viral hepatitis E, non-alcoholic fatty liver disease (NAFLD), alcoholic chronic liver disease, autoimmune hepatitis, primary biliary cirrhosis, hemochromatosis and Wilson disease.
In one embodiment, the patient is a cirrhotic patient.
Another object of the invention is a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices, preferably of large esophageal varices, in a hepatic disease patient, wherein said method comprises:
In one embodiment, the patient was previously diagnosed as cirrhotic, or wherein the patient previously obtained a value between the NPV and the PPV cut-offs in a method as described hereinabove.
The present invention also relates to a microprocessor comprising a computer algorithm carrying out the method as described hereinabove.
In the present invention, the following terms have the following meanings:
The present invention relates to non-invasive methods for assessing the presence and/or severity of varices, selected from esophageal varices and gastric varices in a liver disease patient, preferably is a patient with chronic liver disease.
In one embodiment, the method of the invention is an in vitro method.
In one embodiment, the method of the invention is for assessing the presence of esophageal varices, preferably of esophageal varices of at least about 1 mm in diameter, preferably of at least about 1.5, 2, 2.5, 3, 3.5, 4, 4.5, 5, 6, 7, 8, 9, 10, 15, or 20 mm or more, such as, for example, when measured by UGIE. In one embodiment, the method of the invention is for assessing the presence of large esophageal varices, i.e. of esophageal varices of at least 15% of the esophageal circumference, preferably of at least 25, 30, 40, 50% or more when measured by ECE, or varices of at least about 5 mm in diameter, such as, for example, when measured by UGIE.
In another embodiment, the method of the invention is for assessing the presence of gastric varices such as, for example, gastro-esophageal varices or preferably isolated gastric varices, usually fundal varices.
The present invention first relates to a non-invasive method for assessing the presence and/or severity of varices, selected from esophageal varices and gastric varices in a liver disease patient, using a non-invasive test for assessing the severity of a hepatic lesion or disorder.
The present invention also relates to a non-invasive method for assessing the presence and/or severity of varices, selected from esophageal varices and gastric varices in a liver disease patient, using one or more non-invasive tests for assessing the severity of a hepatic lesion or disorder. The present invention relates to a non-invasive method for assessing the presence and/or severity of varices, selected from esophageal varices and gastric varices in a liver disease patient, using at least two, at least three, at least four or more non-invasive tests for assessing the severity of a hepatic lesion or disorder. In one embodiment, the present invention relates to a non-invasive method for assessing the presence and/or severity of varices, selected from esophageal varices and gastric varices in a liver disease patient, using two, three, four, five or more non-invasive tests for assessing the severity of a hepatic lesion or disorder.
However, in the usual test(s) for assessing the severity of a hepatic lesion or disorder, the cut-offs of said test(s) for assessing the presence and/or severity of varices are determined, preferably as described hereinabove.
Hence, this invention relates to a method comprising:
This invention also relates to a method comprising:
This invention also relates to a method comprising:
In one embodiment, the invention relates to a method comprising carrying out at least two non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
In another embodiment, the invention relates to a method comprising carrying out at least three non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
In another embodiment, the invention relates to a method comprising carrying out at least four non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
In one embodiment, the invention relates to a method comprising carrying out simultaneously two non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
In one embodiment, the invention relates to a method comprising carrying out sequentially two non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices. In another embodiment, the invention relates to a method comprising carrying out sequentially three or more non-invasive tests, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
Hence in one embodiment, the invention relates to a method comprising carrying out two non-invasive tests in an algorithm, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices. In another embodiment, the invention relates to a method comprising carrying out three, or four, or five, or more non-invasive tests in an algorithm, wherein said non-invasive tests each result in a value, said values being compared with cut-offs of said tests for assessing the presence and/or severity of varices.
In one embodiment, the two, three, four, five or more non-invasive tests carried out in the method of the invention are different. Hence in one embodiment, the two, three, four, five or more non-invasive tests carried out in the method of the invention are not repetitions of the same non-invasive tests.
In one embodiment, the method of the invention further comprises a first step of determining the cut-offs of said test(s) for assessing the presence and/or severity of varices, using a population of reference.
Two cut-offs may usually be determined for diagnostic tests, i.e. the NPV cut-off and the PPV cut-off. A value below the NPV cut-off is indicative of the absence of the diagnostic target, whereas a value above the PPV cut-off is indicative of the presence of the diagnostic target. Between the NPV cut-off and the PPV cut-off is an indeterminate zone, wherein no conclusion may be raised regarding the presence or absence of the diagnosis target.
Hence the cut-offs determined for a diagnostic test, the NPV and PPV cut-offs determine two predictive zones: the NPV predictive zone below the NPV cut-off, and the PPV predictive zone above the PPV cut-off. The zone between the NPV and PPV cut-offs is referred to as the indeterminate zone.
In one embodiment, the method of the invention comprises carrying out one non-invasive test for assessing the severity of a hepatic lesion or disorder in step a). Said one non-invasive test is associated with two cut-offs. In one embodiment said cut-offs are NPV and PPV cut-offs thereby defining a NPV predictive zone below the NPV cut-off, and a PPV predictive zone above the PPV cut-off.
In another embodiment, step a) of the method of the invention comprises carrying out two non-invasive tests for assessing the severity of a hepatic lesion or disorder in an algorithm. Said two non-invasive test, for example non-invasive test x and non-invasive test y, are each associated with two cut-offs. In one embodiment each non-invasive test is associated with a NPV and a PPV cut-offs, said NPV (for example NPVx and NPVy) and PPV (for example PPVx and PPVy) cut-offs defining the predictive zones of the algorithm. In one embodiment, the NPV predictive zone is below at least one of the two NPV cut-offs (below NPVx or NPVy) and the PPV predictive zone is above the two PPV cut-offs (above PPVx and PPVy). In another embodiment, the NPV predictive zone is below the two cut offs (below NPVx and NPVy), and the PPV predictive zone is above the two PPV cut-offs (above PPVx and PPVy).
In one embodiment, the method of the invention allows the assessment of the presence and/or severity of varices through the use of single predictive zones, i.e. through the use of one NPV and one PPV predictive zone.
In one embodiment, the method of the invention further comprises, in particular for patients classified in the indeterminate zone between the NPV and PPV cut-offs, one or more repetition of step (a) and step (b) wherein at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out. In another embodiment, the method of the invention further comprises, in particular for patients classified in the indeterminate zone, one or more repetition of step (a) and step (b) wherein the algorithm carried out for assessing the severity of a hepatic lesion or disorder is different from the algorithm previously carried out.
Hence in one embodiment, the NPV and PPV cut-offs determined for the at least one non-invasive test carried out in the repeated step a) define new NPV and PPV predictive zones. In another embodiment, the sets of NPV and PPV cut-offs determined for the at least one non-invasive test carried out in the second step a) and for the at least one non-invasive test carried out in any subsequent step a) each define new NPV and PPV predictive zones. In one embodiment, the method of the invention allows the assessment of the presence and/or severity of varices through the use of multiple predictive zones.
In one embodiment, the method of the invention further comprises a first step of determining the cut-offs of said test(s) for assessing the presence and/or severity of varices, and the associated predictive zones using a population of reference. In another embodiment, the method of the invention further comprises a first step of determining the cut-offs of said test(s) for assessing the presence and/or severity of varices carried out in one or more repetition of step a), and the associated multiple predictive zones using a population of reference.
According to one embodiment, to determine multiple predictive zones in a population of reference, the NPV and PPV predictive zones are first determined using the two non-invasive tests having the largest predictive zones. The choice of the two tests can be done according to several classical statistical techniques, for example the most accurate tests according to multivariate analysis or correlation. The NPV and PPV predictive zones are determined as described hereinabove, using the NPV and PPV cut-offs of each of the two non-invasive tests. Then, a new population of reference is obtained by excluding the patients of the original population of reference located in the NPV and PPV predictive zones. Subsequently new NPV and PPV predictive zones are determined on the smaller population of reference using a different set of two non-invasive tests. At least one of the two non-invasive tests must be different from those used in the first set. Otherwise, the NPV and PPV zones will be empty since the patients within a NPV and PPV zone thus determined have already been excluded. Thus, using the NPV and PPV cut-offs of the new set of two non-invasive tests, new NPV and PPV predictive zones are determined. The process can be reiterated on a new smaller population of reference by excluding the patients located in the second NPV and PPV predictive zones.
In one embodiment, the method of the invention comprises one or more repetition of step a) and step b), wherein at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out.
In another embodiment, the method of the invention comprises two or more repetitions of step a) and step b), wherein for each repetition, at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out.
In another embodiment, the method of the invention comprises three, four, five or more repetitions of step a) and step b), wherein for each repetition, at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out.
In one embodiment, the cut-offs are sensitivity cut-offs and specificity cut-offs. A value below the sensitivity cut-off is indicative of the absence of the diagnostic target, whereas a value above the specificity cut-off is indicative of the presence of the diagnostic target. Between the sensitivity cut-off and the specificity cut-off is an indeterminate zone, wherein no conclusion may be raised regarding the presence or absence of the diagnosis target.
In the present invention, the diagnostic target is the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), and a value below the NPV cut-off is indicative of the absence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), whereas a value above the PPV cut-off is indicative of the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices). Between the NPV cut-off and the PPV cut-off is an indeterminate zone, wherein no conclusion may be raised regarding the presence or absence of varices, selected from gastric or esophageal varices (preferably large esophageal varices).
In one embodiment,
In the present invention, the diagnostic target is the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), and a value below the sensitivity cut-off is indicative of the absence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), whereas a value above the specificity cut-off is indicative of the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices). Between the sensitivity cut-off and the specificity cut-off is an indeterminate zone, wherein no conclusion may be raised regarding the presence or absence of varices, selected from gastric and esophageal varices (preferably large esophageal varices).
The skilled artisan knows how to determine cut-offs for a diagnostic target (see for example a method for determining cut-offs of a diagnostic target in Cales, Liver Intern 2008), using a reference population.
In one embodiment, the NPV cut-offs and the PPV cut-offs are determined in a reference population in order to reach:
In one embodiment, the NPV cut-offs and the PPV cut-offs are determined in a reference population in order to reach a NPV of at least 95% and a PPV of at least 90%.
In one embodiment, the sensitivity cut-offs and the specificity cut-offs are determined in a reference population in order to reach:
In one embodiment, the sensitivity cut-offs and the specificity cut-offs are determined in a reference population in order to reach a sensitivity of at least 95% and a specificity of at least 90%.
In one embodiment, the reference population comprises liver disease patients, preferably patients with chronic liver disease, wherein for each patient the value of the non-invasive test was measured and the status regarding varices, selected from gastric and esophageal varices is known, i.e. absence or presence or size of varices, selected from gastric and esophageal varices, preferably of large esophageal varices (i.e. in one embodiment, an upper gastro-intestinal endoscopy was performed).
Therefore, the present invention is based on the application of a diagnostic test constructed for diagnosing the severity of a hepatic lesion or disorder to the diagnostic of another diagnostic target, varices, through the determination of cut-offs specific for esophageal varices diagnostic.
The experimental data provided in the Examples surprisingly demonstrated that the use of cut-offs specific for varices, selected from gastric and esophageal varices diagnostic instead of cut-offs specific for cirrhosis diagnosis increases the accuracy of the diagnostic test for diagnosing varices, selected from gastric and esophageal varices.
For example, CirrhoMeter™ is a non-invasive diagnostic test primarily constructed for diagnosing cirrhosis (i.e. cut-offs specific for cirrhosis were measured). In the present invention, CirrhoMeter™ cut-offs specific for esophageal varices (preferably large esophageal varices) were measured. CirrhoMeter™ cut-offs for cirrhosis or esophageal varices are shown in the table below.
Therefore, in one embodiment, the method of the invention is for classifying a patient into one of the three following classes:
In one embodiment, the method of the invention further comprises, in particular for patients classified in the indeterminate zone, one or more repetition of step (a) and step (b) wherein at least one non-invasive test carried out for assessing the severity of a hepatic lesion or disorder is different from the at least one non-invasive test previously carried out, thereby defining new NPV and PPV predictive zones and assessing the presence and/or severity of varices in said patient through the use of multiple NPV and PPV predictive zones.
In one embodiment, the method of the invention further comprises, in particular for patients classified in the indeterminate zone, the following steps:
In one embodiment, the assessment of the presence and/or severity of step (f) comprises comparing the score obtained in step (e) with cut-off values for the diagnostic test resulting in the diagnostic score of the invention. As explained hereinabove, two cut-offs may be determined for the diagnostic test resulting in the diagnostic score of the invention: the NPV cut-off and the PPV cut-off, or the sensitivity cut-off and the specificity cut-off.
In one embodiment, the NPV cut-offs and the PPV cut-offs are determined in a reference population in order to reach:
In one embodiment, the NPV cut-offs and the PPV cut-offs are determined in a reference population in order to reach a NPV of at least 95% and a PPV of at least 90%.
In one embodiment, the sensitivity cut-offs and the specificity cut-offs are determined in a reference population in order to reach:
In one embodiment, the sensitivity cut-offs and the specificity cut-offs are determined in a reference population in order to reach a sensitivity of at least 95% and a PPV of at least 90%.
In the present invention, the diagnostic target is the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), and a diagnostic score below the NPV (or sensitivity) cut-off is indicative of the absence of varices, selected from gastric and esophageal varices (preferably large esophageal varices), whereas a diagnostic score above the PPV (or specificity) cut-off is indicative of the presence of varices, selected from gastric and esophageal varices (preferably large esophageal varices). Between the NPV (or sensitivity) cut-off and the PPV (or specificity) cut-off is an indeterminate zone, wherein no conclusion may be raised regarding the presence or absence of varices, selected from gastric and esophageal varices (preferably large esophageal varices).
Therefore, in one embodiment, the method of the invention is for classifying a patient into one of the three following classes:
In one embodiment, patients having a diagnostic score between the NPV and PPV cut-offs required an invasive test for determining the presence or absence of varices, selected from gastric and esophageal varices, such as, for example, endoscopy (UGIE).
In one embodiment, patients having a diagnostic score between the sensitivity and specificity cut-offs required an invasive test for determining the presence or absence of varices, selected from gastric and esophageal varices, such as, for example, endoscopy (UGIE).
In one embodiment, in step (c), the obtained variables are the variables of the non-invasive test carried out in step (a).
In one embodiment, at step (e), the variables obtained at step (c) are mathematically combined in a non-invasive test value, preferably in a score, prior to the mathematical combination with the data obtained at step (d).
In one embodiment, the present invention thus relates to a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices (preferably of large esophageal varices) in a liver disease patient, preferably in a patient with chronic liver disease, wherein said method comprises:
An algorithm corresponding to the non-invasive diagnostic method of the invention is shown in
In one embodiment, the determination of cut-offs for gastric varices is performed in the same way as for large esophageal varices (as illustrated in the Examples): first those of non-invasive test and then those of a score combining non-invasive test and ECE.
In one embodiment, the non-invasive test for assessing the severity of a hepatic lesion or disorder is a biomarker, a clinical data, a binary marker, a blood test or a physical method.
In one embodiment, the non-invasive test results in a value, preferably in a score.
In one embodiment, the non-invasive test is selected from the group comprising age, spleen diameter, ALT, leucocytes, body mass index, GGT, alpha2-macroglobulin, weight, segmented leucocytes, height, monocytes, hemoglobin, P2/MS score, alpha-fetoprotein, alkaline phosphatases, sodium, platelets, AST, InflaMeter, creatinine, urea, APRI, Child-Pugh score, FIB-4, VCTE, albumin, FibroMeter (such as, for example, FibroMeter for cause, FibroMeterV2G or FibroMeterV3G), prothrombin index, CirrhoMeter (such as, for example, CirrhoMeterV2G or CirrhoMeterVV3G), bilirubin, Elasto-FibroMeter (such as, for example, Elasto-FibroMeterV2G), hyaluronate, QuantiMeter (such as, for example, QuantiMeter for cause or QuantiMeterV2G), Hepascore, Fibrotest, Fibrospect, Elasto-Fibrotest, ELF score and any mathematical combination thereof, such as, for example, AST/ALT, AST/ALT+prothrombin, AST/ALT+hyaluronate.
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from platelets, ELF, FibroSpect™, APRI, FIB-4, Hepascore, Fibrotest™, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™; VCTE, ARFI, VTE, supersonic elastometry and/or MRI stiffness.
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from ELF, FibroSpect™, APRI, FIB-4, Hepascore, Fibrotest™, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™; VCTE, ARFI, VTE, supersonic elastometry and/or MRI stiffness.
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™; VCTE, ARFI, VTE, supersonic elastometry and/or MRI stiffness.
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, and/or VCTE (also known as Fibroscan).
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and/or InflaMeter™.
In another embodiment, the at least one non-invasive test carried out in step (a) is selected from FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, InflaMeter™, and/or VCTE (also known as Fibroscan).
Preferably, the at least one non-invasive test carried out in step (a) is selected from FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, and/or InflaMeter™.
In one embodiment, the method of the invention does not comprise carrying out a Fibrotest™.
Examples of biomarkers include, but are not limited to, glycemia, total cholesterol, HDL cholesterol (HDL), LDL cholesterol (LDL), AST (aspartate aminotransferase), ALT (alanine aminotransferase), ferritin, platelets (PLT), prothrombin time (PT) or prothrombin index (PI) or INR (International Normalized Ratio), hyaluronic acid (HA or hyaluronate), haemoglobin, triglycerides, alpha-2 macroglobulin (A2M), gamma-glutamyl transpeptidase (GGT), urea, bilirubin (such as, for example, total bilirubin), apolipoprotein A1 (ApoA1), type III procollagen N-terminal propeptide (P3NP or P3P), gamma-globulins (GBL), sodium (Na), albumin (ALB) (such as, for example, serum albumin), ferritine (Fer), glucose (Glu), alkaline phosphatases (ALP), YKL-40 (human cartilage glycoprotein 39), tissue inhibitor of matrix metalloproteinase 1 (TIMP-1), TGF, cytokeratine 18 and matrix metalloproteinase 2 (MMP-2) to 9 (MMP-9), haptoglobin, alpha-fetoprotein, creatinine, leukocytes, neutrophils, segmented leukocytes, segmented neutrophils, monocytes, ratios and mathematical combinations thereof, such as, for example AST/ALT (ratio), AST.ALT (product), AST/PLT (ratio), AST/ALT+prothrombin, AST/ALT+hyaluronate.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring platelets (PLT).
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder and optionally measuring the platelet count in a blood sample from said patient, wherein said at least one non-invasive test and optionally said platelet count result in at least one value.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder and measuring the platelet count in a blood sample from said patient, wherein said at least one non-invasive test and said platelet count each result in at least one value.
The measurements carried out in the method of the invention are measurements aimed either at quantifying the biomarker (such as, for example, in the case of A2M, HA, bilirubin, PLT, PT, urea, NA, glycemia, triglycerides, ALB or P3P), or at quantifying the enzymatic activity of the biomarker (such as, for example, in the case of GGT, ASAT, ALAT, ALP). Those skilled in the art are aware of various direct or indirect methods for quantifying a given substance or a protein or its enzymatic activity. These methods may use one or more monoclonal or polyclonal antibodies that recognize said protein in immunoassay techniques (such as, for example, radioimmunoassay or RIA, ELISA assays, Western blot, etc.), the analysis of the amounts of mRNA for said protein using techniques of the Northern blot, slot blot or PCR type, techniques such as an HPLC optionally combined with mass spectrometry, etc. The abovementioned protein activity assays use assays carried out on at least one substrate specific for each of these proteins. International patent application WO 03/073822 lists methods that can be used to quantify alpha2 macroglobulin (A2M) and hyaluronic acid (HA or hyaluronate).
By way of examples, and in a non-exhaustive manner, a preferred list of commercial kits or assays that can be used for the measurements of biomarkers carried out in the method of the invention, on blood samples, is given hereinafter:
For the biomarkers measured in the method of the present invention, the values obtained may be expressed in:
Examples of clinical data include, but are not limited to, weight, height, body mass index, age, sex, hip perimeter, abdominal perimeter or height, spleen diameter (preferably by abdominal imaging), and mathematical combinations thereof, such as, for example, the ratio thereof, such as for example hip perimeter/abdominal perimeter.
Examples of non-invasive binary markers include, but are not limited to, diabetes, SVR (wherein SVR stands for sustained virologic response, and is defined as aviremia 6 weeks, preferably 12 weeks, more preferably 24 weeks after completion of antiviral therapy for chronic hepatitis C virus (HCV) infection), etiology, hepatic encephalopathy, ascites, and NAFLD. Regarding the binary marker “etiology”, the skilled artisan knows that said variable is a single or multiple binary marker, and that for liver disorders, etiology may be NAFLD, alcohol, virus or other. Thus, the binary marker might be expressed as NAFLD vs others (single binary marker) or as NAFLD vs reference etiology plus virus vs reference etiology and so on (multiple binary marker).
Preferably, the data is an elastometry data, preferably Liver Stiffness Evaluation (LSE) data or spleen stiffness evaluation, which may be for example obtained by VCTE or ARFI or SSI or another elastometry technique. According to a preferred embodiment of the invention, the physical data is liver stiffness measurement (LSM), preferably measured by VCTE.
In a particular embodiment, the physical data is Liver stiffness measurement (LSM) by VCTE (also known as Fibroscan™, Paris, France), preferably performed with the M probe. Preferably, examination conditions are those recommended by the manufacturer, with the objective of obtaining at least 3 and preferably 10 valid measurements. Results may be expressed as the median (kilopascals) of all valid measurements, or as IQR or as the ratio (IQR/median).
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out a VCTE (also known as Fibroscan™).
In one embodiment, step (a) of the non-invasive method of the invention comprises obtaining a liver stiffness measurement (LSM) by VCTE (also known as Fibroscan™).
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out a VCTE (also known as Fibroscan™) and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out a VCTE (also known as Fibroscan™) and measuring the platelet count in a blood sample from said patient.
In one embodiment, the realization of a VCTE (also known as Fibroscan™) and the measurement of the platelet count and the comparison of the values obtained with cut-offs for assessing the presence and/or severity of varices corresponds to a PlFS algorithm.
Example 4 provides examples of PlFS algorithms.
In one embodiment, the blood test of the invention corresponds to a blood test selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, Fibrotest™, FibroMeter™ (such as, for example, FibroMeter for cause, FibroMeterV2G or FibroMeterV3G), CirrhoMeter™ (such as, for example, CirrhoMeterV2G or CirrhoMeterV3G), CombiMeter, Elasto-FibroMeter™ (such as, for example, Elasto-FibroMeterV2G), InflaMeter™, Actitest, QuantiMeter, P2/MS score, Elasto-Fibrotest, and Child-Pugh score. As these blood tests are diagnostic tests, they can be based on multivariate mathematical combination, such as, for example, binary logistic regression, or include clinical data.
ELF is a blood test based on hyaluronic acid, P3P, TIMP-1 and age.
FibroSpect™ is a blood test based on hyaluronic acid, TIMP-1 and A2M.
APRI is a blood test based on platelet and AST.
FIB-4 is a blood test based on platelet, AST, ALT and age.
HEPASCORE is a blood test based on hyaluronic acid, bilirubin, alpha2-macroglobulin, GGT, age and sex.
FIBROTEST™ is a blood test based on alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, age and sex.
FIBROMETER™ and CIRRHOMETER™ together form a family of blood tests, the content of which depends on the cause of chronic liver disease and the diagnostic target (such as, for example, fibrosis, significant fibrosis or cirrhosis). This blood test family is called FM family and is detailed in the table below.
aHA is replaced by GGT
COMBIMETER™ or Elasto-FibroMeter™ is a family of tests based on the mathematical combination of variables of the FM family (as detailed in the Table hereinabove) or of the result of a test of the FM family with VCTE (FIBROSCAN™) result. In one embodiment, said mathematical combination is a binary logistic regression.
In one embodiment, CombiMeter™ or Elasto-FibroMeter™ results in a score based on the mathematical combination of physical data from liver or spleen elastometry such as dispersion index from VCTE (Fibroscan™) such as IQR or IQR/median or median of LSM, preferably of LSM (by Fibroscan™) median with at least 3, 4, 5, 6, 7, 8 or 9 biomarkers and/or clinical data selected from the list comprising glycemia, total cholesterol, HDL cholesterol (HDL), LDL cholesterol (LDL), AST (aspartate aminotransferase), ALT (alanine aminotransferase), AST/ALT, AST.ALT, ferritin, platelets (PLT), AST/PLT, prothrombin time (PT) or prothrombin index (PI), hyaluronic acid (HA or hyaluronate), haemoglobin, triglycerides, alpha-2 macroglobulin (A2M), gamma-glutamyl transpeptidase (GGT), urea, bilirubin, apolipoprotein A1 (ApoA1), type III procollagen N-terminal propeptide (P3NP), gamma-globulins (GBL), sodium (Na), albumin (ALB), ferritine (Fer), glucose (Glu), alkaline phosphatases (ALP), YKL-40 (human cartilage glycoprotein 39), tissue inhibitor of matrix metalloproteinase 1 (TIMP-1), TGF, cytokeratine 18 and matrix metalloproteinase 2 (MMP-2) to 9 (MMP-9), haptoglobin, diabetes, weight, body mass index, age, sex, hip perimeter, abdominal perimeter or height and ratios and mathematical combinations thereof.
INFLAMETER™ is a companion test reflecting necro-inflammatory activity including ALT, A2M, PI, and platelets.
ACTITEST is a blood test based on alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, ALT, age and sex.
QUANTIMETER is a blood test based on (i) alpha2-macroglobulin, hyaluronic acid, prothrombin time, platelets when designed for alcoholic liver diseases, (ii) hyaluronic acid, prothrombin time, platelets, AST, ALT and glycemia when designed for NAFLD, or (iii) alpha2-macroglobulin, hyaluronic acid, platelets, urea, GGT and bilirubin when designed for chronic viral hepatitis.
P2/MS is a blood test based on platelet count, monocyte fraction and segmented neutrophil fraction.
CHILD-PUGH SCORE is a blood test based on total bilirubin, serum albumin, PT or INR, ascites and hepatic encephalopathy.
ELASTO-FIBROTEST is a test based on the mathematical combination of variables of FIBROTEST or of the result of a FIBROTEST, with LSM measurement, measured for example by Fibroscan™.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of ELF, i.e. hyaluronic acid, P3P, TIMP-1 and age.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FibroSpect™, i.e. hyaluronic acid, TIMP-1 and A2M.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of APRI, i.e. platelet and AST.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FIB-4, i.e. platelet, AST, ALT and age.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of HEPASCORE, i.e. hyaluronic acid, bilirubin, alpha2-macroglobulin, GGT, age and sex.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FIBROTES™, i.e. alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, age and sex.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FIBROMETER™ and/or CIRRHOMETER™ as defined hereinabove.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FIBROMETER™ and/or CIRRHOMETER™ as defined hereinabove and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of FIBROMETER™ and/or CIRRHOMETER™ as defined hereinabove and measuring the platelet count in a blood sample from said patient.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, i.e. the following variables:
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, i.e. the following variables:
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, i.e. the following variables:
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of COMBIMETER™ or Elasto-FibroMeter™ as defined hereinabove.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of INFLAMETER™, i.e. ALT, A2M, PI, and platelets.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of ACTITEST, i.e. alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, ALT, age and sex.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of QUANTIMETER, i.e. (i) alpha2-macroglobulin, hyaluronic acid, prothrombin time, platelets, (ii) hyaluronic acid, prothrombin time, platelets, AST, ALT and glycemia, or (iii) alpha2-macroglobulin, hyaluronic acid, platelets, urea, GGT and bilirubin.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of P2/MS score, i.e. platelet count, monocyte fraction and segmented neutrophil fraction.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CHILD-PUGH SCORE, i.e. total bilirubin, serum albumin, PT or INR, ascites and hepatic encephalopathy.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least two non-invasive tests for assessing the severity of a hepatic lesion or disorder, wherein said at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least two non-invasive tests for assessing the severity of a hepatic lesion or disorder and optionally measuring the platelet count in a blood sample from said patient, wherein said at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least two non-invasive tests for assessing the severity of a hepatic lesion or disorder and measuring the platelet count in a blood sample from said patient, wherein said at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™ and VCTE (also known as Fibroscan™); and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™ and VCTE (also known as Fibroscan™); and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness, and optionally measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™ and VCTE (also known as Fibroscan™); and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness, and measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness and optionally measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising ELF, FibroSpect™, APRI, FIB-4, Hepascore, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness and measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness, wherein said the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness and optionally measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises carrying out at least one non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, and InflaMeter™; and another non-invasive test for assessing the severity of a hepatic lesion or disorder selected from the group comprising, FibroMeter™, CirrhoMeter™, CombiMeter, Elasto-FibroMeter™, Elasto-Fibrotest, InflaMeter™, VCTE (also known as Fibroscan™), ARFI, VTE, supersonic elastometry and MRI stiffness and measuring the platelet count in a blood sample from said patient, wherein the at least two non-invasive tests are different.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, and measuring and combining in a mathematical function the variables of FIBROMETER™.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, and measuring and combining in a mathematical function the variables of FIBROMETER™, and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, and measuring and combining in a mathematical function the variables of FIBROMETER™, and measuring the platelet count in a blood sample from said patient.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a FibroMeter.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a FibroMeter and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a FibroMeter and measuring the platelet count in a blood sample from said patient.
Hence in one embodiment, the non-invasive method of the invention comprises:
In one embodiment, the realization of a CirrhoMeter and a FibroMeter and the comparison of the values obtained with cut-offs of CirrhoMeter and FibroMeter for assessing the presence and/or severity of varices corresponds to a CMFM algorithm.
Examples 3 and 4 provide examples of CMFM algorithms.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, and obtaining a liver stiffness measurement (LSM) by VCTE (also known as Fibroscan™).
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, obtaining a liver stiffness measurement (LSM) by VCTE (also known as Fibroscan™), and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, step (a) of the non-invasive method of the invention comprises measuring and combining in a mathematical function the variables of CIRRHOMETER™, obtaining a liver stiffness measurement (LSM) by VCTE (also known as Fibroscan™), and measuring the platelet count in a blood sample from said patient.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a VCTE (also known as Fibroscan™).
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a VCTE (also known as Fibroscan™) and optionally measuring the platelet count in a blood sample from said patient.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter and a VCTE (also known as Fibroscan™) and measuring the platelet count in a blood sample from said patient.
Hence in one embodiment, the non-invasive method of the invention comprises:
In one embodiment, the realization of a CirrhoMeter and a VCTE (also known as Fibroscan™) and the comparison of the values obtained with cut-offs of CirrhoMeter and VCTE for assessing the presence and/or severity of varices corresponds to a CMFS algorithm.
Examples 2 and 4 provide examples of CMFS algorithms, including the CMFS#1 algorithm.
In one embodiment, the realization of a CirrhoMeter and a VCTE (also known as Fibroscan™) and the comparison of the values obtained with cut-offs of CirrhoMeter and VCTE for assessing the presence and/or severity of varices corresponds to the algorithm CMFS#1.
In one embodiment, the non-invasive method of the invention comprises carrying out the CMSF#1 algorithm.
In one embodiment, the realization of a CirrhoMeter and a VCTE (also known as Fibroscan™), the measurement of the platelet count and the comparison of the values obtained with cut-offs for assessing the presence and/or severity of varices corresponds to a PlCMFS algorithm.
Example 4 provides an example of PlCMFS algorithm.
In one embodiment, the method of the invention comprises carrying out a CirrhoMeter, a FibroMeter and a VCTE (also known as Fibroscan™), and measuring the platelet count.
In one embodiment, the realization of a CirrhoMeter, a FibroMeter and a VCTE (also known as Fibroscan™), the measurement of the platelet count and the comparison of the values obtained with cut-offs for assessing the presence and/or severity of varices corresponds to a PlFMCMFS algorithm.
Example 4 provides an example of PlFMCMFS algorithm.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of ELF, i.e. hyaluronic acid, P3P, TIMP-1 and age.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of FibroSpect™, i.e. hyaluronic acid, TIMP-1 and A2M.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of APRI, i.e. platelet and AST.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of FIB-4, i.e. platelet, AST, ALT and age.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of HEPASCORE, i.e. hyaluronic acid, bilirubin, alpha2-macroglobulin, GGT, age and sex.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of FIBROTEST™, i.e. alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, age and sex.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of FIBROMETER™ and/or CIRRHOMETER™ as defined hereinabove.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of CIRRHOMETER™, i.e. the following variables:
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of COMBIMETER™ or Elasto-FibroMeter™ as defined hereinabove.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of INFLAMETER™, i.e. ALT, A2M, PI, and platelets.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of ACTITEST, i.e. alpha2-macroglobulin, haptoglobin, apolipoprotein A1, total bilirubin, GGT, ALT, age and sex.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of QUANTIMETER, i.e. (i) alpha2-macroglobulin, hyaluronic acid, prothrombin time, platelets, (ii) hyaluronic acid, prothrombin time, platelets, AST, ALT and glycemia, or (iii) alpha2-macroglobulin, hyaluronic acid, platelets, urea, GGT and bilirubin.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of P2/MS score, i.e. platelet count, monocyte fraction and segmented neutrophil fraction.
In one embodiment, step (c) of the non-invasive method of the invention comprises measuring the variables of CHILD-PUGH SCORE, i.e. total bilirubin, serum albumin, PT or INR, ascites and hepatic encephalopathy.
Examples of physical methods include, but are not limited to, medical imaging data and clinical measurements, such as, for example, measurement of spleen, especially spleen length (that may also be referred as diameter). According to an embodiment, the physical method is selected from the group comprising ultrasonography, especially Doppler-ultrasonography and elastometry ultrasonography and velocimetry ultrasonography (preferred tests using said data are vibration controlled transient elastography (VCTE, also known as Fibroscan™), ARFI, VTE, supersonic elastometry (supersonic imaging), MRI (Magnetic Resonance Imaging), and MNR (Magnetic Nuclear Resonance) as used in spectroscopy, especially MNR elastometry or velocimetry.
In one embodiment, the physical method is VCTE, ARFI, VTE, supersonic elastometry or MRI stiffness.
In one embodiment of the invention, the method of the invention comprises carrying out a VCTE, which refers to obtaining at least 3 and preferably 10 valid measurements and recovering a physical data corresponding to the median in kilopascals of all valid measurements.
In one embodiment, the present invention non-invasive method for assessing the presence and/or severity of esophageal varices in a hepatic disease patient comprises:
In one embodiment, the present invention non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices in a hepatic disease patient comprises:
Examples of non-invasive imaging data allowing the assessment of varices status (i.e. for visualizing varices or the absence of varices) include data obtained with non-invasive imaging methods or radiology.
Examples of non-invasive imaging methods for assessing varices status include, but are not limited to, esophageal capsule endoscopy (ECE), CT-scan, echo-endoscopy or MRI.
Examples of esophageal capsules that may be used in the method of the present invention includes esophageal capsules developed by Given-covidien-medtronic.
Examples of radiologic methods for assessing varices status include, but are not limited to, CT-scanner and MRI.
In one embodiment, the non-invasive imaging data corresponds to a grade according to the size of the visualized varices:
In one embodiment, the step (a) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G.
In one embodiment, the step (c) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G.
In one embodiment, the step (a) and step (c) of the method of the invention both comprise carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeteV3G, preferably a CirrhoMeterV2G.
In one embodiment, the step (a) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G.
In one embodiment, the step (c) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G.
In one embodiment, the step (a) and step (c) of the method of the invention both comprise carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G.
In one embodiment, the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE.
In one embodiment, the step (e) of the method of the invention comprises mathematically combining a CirrhoMeter (such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G) or the variables of a CirrhoMeter (such as, for example, a CirrhoMeterV2G or a CirrhoMeterVV3G, preferably a CirrhoMeterV2G) with a data obtained by ECE.
In one embodiment, the step (e) of the method of the invention comprises mathematically combining a FibroMeter (such as, for example, a FibroMeterV2G or a FibroMeterV3G) or the variables of a FibroMeter (such as, for example, a FibroMeterV2G or a FibroMeterV3G) with a data obtained by ECE.
In one embodiment, the step (c) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the CirrhoMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the step (c) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the FibroMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the step (a) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G; the step (c) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the CirrhoMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the step (a) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G; the step (c) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the FibroMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the step (a) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G; the step (c) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the CirrhoMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the step (a) of the method of the invention comprises carrying out a CirrhoMeter, such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G; the step (c) of the method of the invention comprises carrying out a FibroMeter, such as, for example, a FibroMeterV2G or a FibroMeterV3G; the step (d) of the method of the invention comprises obtaining imaging data obtained by ECE; and the step (e) of the method of the invention comprises mathematically combining the result of the FibroMeter carried out at step (c) with the data obtained by ECE.
In one embodiment, the patient is a mammal, preferably a human. In one embodiment, the patient is a male or a female. In one embodiment, the patient is an adult or a child.
In one embodiment, the patient is affected, preferably is diagnosed with a liver disease or disorder.
In one embodiment, the patient is affected with a liver disease or disorder, preferably selected from the list comprising significant porto-septal fibrosis, severe porto-septal fibrosis, centrolobular fibrosis, cirrhosis, persinusoidal fibrosis, the fibrosis being from alcoholic or non-alcoholic origin.
In one embodiment, the patient is affected with a chronic disease, preferably said chronic disease is selected from the group comprising chronic viral hepatitis C, chronic viral hepatitis B, chronic viral hepatitis D, chronic viral hepatitis E, non-alcoholic fatty liver disease (NAFLD), alcoholic chronic liver disease, autoimmune hepatitis, primary biliary cirrhosis, hemochromatosis and Wilson disease.
In one embodiment, the subject is a cirrhotic patient. In one embodiment, the patient was previously diagnosed as cirrhotic by any method known in the art, including invasive (e.g. biopsy) or non-invasive (e.g. blood test or physical method) methods already disclosed in the art.
In one embodiment of the invention, the mathematical combination is a combination within a mathematical function selected from a binary logistic regression, a multiple linear regression or any multivariate analysis. One skilled in the art may found in the prior art all information related to the mathematical function.
In one embodiment, the mathematical function is a logistic regression. A logistic regression produces a formula in the form:
score=a0+a1x1+a2x2+ . . .
wherein the coefficients ai are constants and the variables xi are the variables (preferably independent variables).
Preferably, the mathematical function is a binary logistic regression where final score is 1/1-escore.
In one embodiment, the diagnostic method of the invention presents:
In one embodiment, the diagnostic method of the invention presents a NPV of at least 95% and/or a PPV of at least 90%.
In one embodiment, the diagnostic method of the invention presents a diagnostic performance (patients correctly classified or AUROC) for esophageal varices, preferably for large esophageal varices, of at least about 0.89, preferably of at least about 0.90, 0.91, 0.92, 0.93, 0.94, 0.95, 0.96, 0.97, 0.98, 0.99 or more.
In one embodiment, the percentage of correctly classified patients using the method of the invention is of at least about 90%, preferably of at least about 90.5, 91, 91.5, 92, 92.5, 93, 93.5, 94, 94.5, 95, 95.5, 96, 96.5, 97, 97.5, 98, 98.5, 99, 99.5 or more.
In one embodiment, the diagnostic method of the invention presents a specificity of at least 90%, preferably of at least 91, 92, 93, 94, 95, 96, 97, 98, 99% or more. In one embodiment, the diagnostic method of the invention presents a specificity of 100%.
In one embodiment, using the diagnostic method of the invention, an invasive test for determining the presence or absence of esophageal varices, such as, for example, endoscopy (UGIE) is required in at most about 50%, preferably in at most about 45, 40, 35, 30, 25, 20, 15, 10% or less of the hepatic disease patients.
In one embodiment, using the diagnostic method of the invention, the rate of saved UGIE is of at least about 20%, preferably of at least about 30, 40, 50, 60, 70, 80, 90% or more.
In one embodiment, using the diagnostic method of the invention, the rate of missed large esophageal varices is of at most about 20%, preferably of at most about 19, 18, 17, 16, 15, 14, 13, 12, 11, 10, 9, 8, 7, 6, 5, 4, 3, 2, 1% or less.
Another object of the invention is a non-invasive method for assessing the presence and/or severity of varices, selected from gastric and esophageal varices in a liver disease patient, preferably in a patient with chronic liver disease, wherein said method comprises:
In one embodiment, the biomarkers, clinical data, binary markers, physical data and imaging data on varices status are as defined hereinabove.
In one embodiment, the variables measured in step (i) are the variables of a CirrhoMeter (such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G).
In another embodiment, the variables measured in step (i) are the variables of a FibroMeter (such as, for example, a FibroMeterV2G or a FibroMeterV3G).
In one embodiment, the imaging data are obtained in step (ii) by ECE.
In one embodiment, the variables obtained in step (i) are mathematically combined in a non-invasive diagnostic test, preferably in a score, prior to the mathematical combination with the data obtained at step (ii). In one embodiment, the variables obtained in step (i) are mathematically combined in a FibroMeter or in a CirrhoMeter.
In one embodiment, the step (iii) of the method of the invention comprises mathematically combining a CirrhoMeter (such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G) or the variables of a CirrhoMeter (such as, for example, a CirrhoMeterV2G or a CirrhoMeterV3G, preferably a CirrhoMeterV2G) with a data obtained by ECE.
In one embodiment, the step (iii) of the method of the invention comprises mathematically combining a FibroMeter (such as, for example, a FibroMeterV2G or a FibroMeterV3G) or the variables of a FibroMeter (such as, for example, a FibroMeterV2G or a FibroMeterV3G) with a data obtained by ECE.
In one embodiment, the patient was previously diagnosed with a cirrhosis.
In one embodiment, the patient was classified in the indeterminate zone according to the step (b) of the method as defined hereinabove.
In one embodiment of the invention, the method of the invention is computer implemented.
The present invention thus also relates to a microprocessor comprising a computer algorithm carrying out the prognostic method of the invention.
The skilled artisan would easily deduce that the method of the invention being indicative of the presence of varices, selected from gastric and esophageal varices, especially of large esophageal varices, it may be used by the physician willing to provide the best medical care to his/her patient. For example, a patient presenting varices, selected from gastric and esophageal varices will require treatment of said varices, while a patient without esophageal varices will be subjected to yearly surveillance of varices. On the other hand, a patient diagnosed in the indeterminate zone regarding the value of the diagnostic score of the invention may require an UGIE endoscopy in order to assess the presence or absence of varices.
Therefore, the present invention also relates to a method for adapting the treatment, the medical care or the follow-up of a patient, wherein said method comprises implementing the non-invasive method of the invention.
The present invention also relates to a method for monitoring the treatment of a patient, wherein said method comprises implementing the non-invasive method of the invention, thereby assessing the appearance of esophageal varices in a patient.
The present invention also relates to a method for treating a hepatic disease patient, wherein said method comprises (i) implementing the non-invasive method of the invention and (ii) treating the patient according to the value obtained by the patient.
The present invention is further illustrated by the following examples.
Diagnostic strategy development needed a derivation population where all diagnostic tests were available, especially esophageal capsule endoscopy (ECE). For that purpose, we disposed of a population with cirrhotic patients. The derivation population was extracted from a prospective study comparing ECE and upper gastro-intestinal endoscopy (UGIE) in the esophageal varices (EV) diagnosis in patients with cirrhosis with various causes (Sacher-Huvelin S, et al, Endoscopy 2015: (in press: PMID: 25730284)). We included the 287 patients having both ECE and UGIE.
Validation populations were already published for the evaluation of non-invasive fibrosis tests (except population #4). The two main differences with derivation population were (i) the availability of liver biopsy in all patients and (ii) that all fibrosis stages were represented.
Validation of diagnostic strategy required a chronic liver disease (CLD) population with UGIE for reference and blood tests. Briefly, this validation population #1 was particular for several reasons. Patients with CLD attributed to virus or alcohol could have decompensated cirrhosis and had UGIE even in non-cirrhotic patients (Oberti F et al, Gastrointest Endosc 1998; 48:148-157). Finally, we considered 3 additional large validation CLD populations #2 to #4 to mainly validate specificity robustness. Validation populations #2 (Pascal J P et al, N Engl J Med 1987; 317:856-861) and #3 (Castera L et al, J Hepatol 2008; 48:835-847) comprised patients with CLD due to chronic hepatitis C (CHC) without liver complication.
Population #4 comprised patients with CLD due to non-alcoholic fatty liver disease (NAFLD) without liver complication. Patients with biopsy-proven NAFLD were consecutively included in the study from January 2004 to June 2014 at Angers University Hospital and from October 2003 to April 2014 at Bordeaux University Hospital. NAFLD was defined as liver steatosis on liver biopsy after exclusion of concomitant steatosis-inducing drugs, excessive alcohol consumption (>210 g/week in men or >140 g/week in women), chronic hepatitis B or C infection, and histological evidence of other concomitant chronic liver disease. Patients were excluded if they had cirrhosis complications (ascites, variceal bleeding, systemic infection, or hepatocellular carcinoma). The study protocol conformed to the ethical guidelines of the current Declaration of Helsinki and all patients gave informed written consent.
Study design is summarized in table 1 and
The diagnostic algorithms included different strategies (
The first ones comprised a single diagnostic test. The second ones combined several tests. These combinations were either symmetric, i.e. the same test combination for both predictive values (PV), or asymmetric to reach higher PV, i.e. with different tests for negative predictive value (NPV) and positive predictive value (PPV). We defined a clinically applicable strategy as including necessarily a low constraint test (e.g. blood test) to exclude LEV (usually in asymptomatic patient) and possibly a high constraint test (e.g. UGIE) to affirm LEV (in the most severe patients).
First step—We evaluated available strategies combining one or several tests to determine the most accurate strategy in the derivation population irrespective of clinical applicability in order to determine the most accurate strategy as paradigm.
Second step—We concentrated on the sole clinically applicable asymmetric strategies. When we compared strategies, we had no single comparator and a choice had to be based on a balance between the best three indicators (patient proportion with PV, saved UGIE and missed LEV, see below) according to statistical comparisons.
Endoscopic procedures of derivation population are detailed in our previous publication (Sacher-Huvelin S et al, Endoscopy 2015). In validation population #1, UGIE was performed by two senior endoscopists experienced in studies on PHT (Oberti F et al, Gastrointest Endosc 1998; 48:148-157). In this study, EV size was also classified qualitatively into 3 grades: 1: small, 2: medium or 3: large. Stages 2 and 3 were grouped in LEV (Pascal J P et al, N Engl J Med 1987; 317:856-861).
Blood tests—The following blood tests were calculated according to published or patented formulas. Hepascore (Adams L A et al, Clin Chem 2005; 51:1867-1873), Fib-4 (Sterling R K et al, Hepatology 2006; 43:1317-1325), APRI (Wai C T et al, Hepatology 2003; 38:518-526). FibroMeterVIRUS2G (Leroy V et al, Clin Biochem 2008; 41:1368-1376), CirrhoMeterVIRUS2G (Boursier J et al, Eur J Gastroenterol Hepatol 2009; 21:28-38), FibroMeterVIRUS3G (Calès P et al, J Hepatol 2010; 52: S406) and CirrhoMeterVIRUS3G (Calès P et al, J Hepatol 2010; 52: S406) were constructed for Metavir fibrosis staging in CHC. In FibroMeter/CirrhoMeterVIRUS3G GGT replaces hyaluronate included in FibroMeter/CirrhoMeterVIRUS2G. CirrhoMeter tests were constructed for cirrhosis diagnosis and included all FibroMeter markers ((Boursier J et al, Eur J Gastroenterol Hepatol 2009; 21:28-38). FibroMeterALD (Cales P et al, Gastroenterol Clin Biol 2008; 32:40-51) and FibroMeterNAFLD (Cales P et al, J Hepatol 2009; 50:165-173) were constructed for Metavir fibrosis staging, respectively in alcoholic liver disease (ALD) and NAFLD. QuantiMeterNAFLD was constructed to evaluate the area of whole fibrosis in NAFLD (Cales P et al, Liver Int 2010; 30:1346-1354). QuantiMeterVIRUS and QuantiMeterALD were constructed to evaluate the area of whole fibrosis in CHC and ALD, respectively (Cales P, et al, Hepatology 2005; 42:1373-1381). All blood assays were performed in the same laboratories of each center, or partially centralized in population #3. Tests were used as raw data without correction rules like expert system.
Elastometry—Vibration control transient elastography (VCTE) (Fibroscan™, Echosens, Paris, France) examination was performed by an experienced observer (>50 examinations before the study), blinded for patient data. Examination conditions were those recommended by the manufacturer (Castera L et al, J Hepatol 2008; 48:835-847). VCTE examination was stopped when 10 valid measurements were recorded. Results (kilopascals) were expressed as the median and the interquartile range of all valid measurements.
Combined test—One test combined markers of blood test and VCTE: Elasto-FibroMeter2G (E-FibroMeter2G) (Cales P et al, Liver international: official journal of the International Association for the Study of the Liver 2014; 34:907-917).
For LEV diagnosis, we calculated diagnostic indices as a function of test values (
We determined the cut-off of test value to reach NPV≥95% and a PPV≥90% in the largest subpopulation when possible. PPV and NPV were reported through two statistical descriptors. First, the patient proportion (% out of the whole population) being included between the first test value reaching the expected predefined cut-off (95 or 90%) and the extreme test value, called PV patient proportion thereafter (see
Ugie requirement—This is the patient proportion in the indeterminate zone between NPV and PPV cut-offs for LEV.
Missed LEV—This is the proportion of LEV in the NPV zone for LEV.
Saved UGIE—The reference patient group to calculate saved UGIE is the cirrhosis group where UGIE is classically performed: the whole population in derivation population and patients with Metavir F4 stage in validation population #1. The saved UGIE rate is the patient proportion provided by the difference between the reference group and the target group where UGIE is indicated by non-invasive tests. The target group can be determined by cut-offs of fibrosis staging or LEV diagnosis.
Quantitative variables were expressed as mean±standard deviation. 95% confidence intervals (CI) were calculated by bootstrapping on 1000 samples. The discriminative ability of each test was expressed as the area under the receiver operating characteristic (AUROC) curve and compared by the Delong test. Data were reported according to STARD (Bossuyt P M et al, Clin Chem 2003; 49:7-189) and Liver FibroSTARD (Boursier J et al, J Hepatol 2015) statements, and analyzed on an intention to diagnose basis. Scores including independent predictors of LEV were determined by binary logistic regression. In the population where test is constructed, its accuracy is maximized and thus includes an optimism bias. Therefore, this bias was noticed when present. The main statistical analyses were performed under the control of professional statisticians using SPSS version 18.0 (IBM, Armonk, N.Y., USA) and SAS 9.2 (SAS Institute Inc., Cary, N.C., USA).
Characteristics of main populations are described in table 2 and those of ancillary validation populations in table 3.
aEstimation
bIn cirrhosis in brackets
Differences between populations were observed with respect to etiology and severity of liver disease and also concerning the investigations performed (table 1). In validation CLD population #1, LEV were only observed in patients with confirmed cirrhosis according to liver biopsy and in those with probable cirrhosis according to CirrhoMeterVIRUS2G (
Accuracies by AUROC of predictors for LEV are detailed in table 4.
Spleen diameter
119
0.518
0.508
0.697
ALT
287
0.532
0.516
0.705
Leucocytes
283
0.527
0.522
Body mass index
270
0.479
0.526
0.630
GGT
287
0.582
0.526
0.562
Alpha2-macroglobulin
248
0.524
0.529
0.656
Weight
278
0.491
0.531
0.592
Segmented leucocytes
216
0.578
0.534
10.
Height
273
0.568
0.563
0.403
11.
Monocytes
216
0.565
0.571
22.
APRI
284
0.655
0.704
0.682
23.
Child-Pugh score
287
0.718
0.718
0.782
24.
Fib-4
284
0.702
0.725
0.790
25.
VCTE
211
0.738
0.730
26.
Albumin
275
0.727
0.734
0.743
27.
FibroMeter for cause
251
0.736
0.736
28.
AST/ALT
287
0.737
0.747
0.678
29.
Prothrombin index
284
0.733
0.752
30.
FibroMeter
VIRUS3G
243
0.755
0.761
31.
CirrhoMeter
VIRUS3G
243
0.752
0.763
32.
Bilirubin
284
0.738
0.771
33.
Elasto-FibroMeterVIRUS2G
160
0.775
0.773
34.
AST/ALT + prothrombin
284
0.763
0.778
35.
Hyaluronate
225
0.772
0.794
36.
AST/ALT + hyaluronate
225
0.777
0.794
37.
QuantiMeter
VIRUS
210
0.707
0.799
38.
QuantiMeter for cause
249
0.770
0.799
0.800
0.911
0.801
0.863
0.810
0.884
0.874
0.845
0.885
0.867
44.
ECE + CirrhoMeter
VIRUS2G
211
45.
ECE + AST/ALT
287
a 158 patients.
b Validation population #1 including 165 patients
Briefly, in derivation population, the highest AUROC at 0.92 was obtained with ECE+(AST/ALT) score. This score and ECE+CirrhoMeterVIRUS2G score had significantly higher AUROC than fibrosis tests (p<0.02) whereas AUROCs between other fibrosis tests were not significantly different (data not shown). In validation population, AUROC of Metavir F stage was significantly inferior to that of the most accurate tests (CirrhoMeters and FibroMeterVIRUSS2G). This suggests that non-invasive testing can be more effective for LEV diagnosis than a histological diagnosis of cirrhosis.
Derivation population—CirrhoMeterVIRUS2G was the most accurate low constraint test resulting in the highest NPV patient proportion and the highest measured NPV for LEV among all strategies (table 5).
11.4
10.0
96.8
52.5
0
c
20.0
93.0
78.5
11.4
26.7
100
91.7
a Patients diagnosed with small EV by ECE
b It was not possible to reach the objective (≥90%) since this is semi-quantitative variable.
c Maximum PPV was <40%
d This figure is different from ECE alone since 3 patients with high LEV PPV with ECE had high LEV NPV with VCTE
e This figure is different from ECE + AST/ALT score alone (simultaneous combination) since 2 patients with high LEV PPV with ECE + AST/ALT score had high LEV NPV with VCTE
f By paired Cochran test for patient proportions. Useful pairwise comparisons are provided in the text
In the largest sample size tested for CirrhoMeterVIRUS2G (table 6) the measured PPV was 98% (95% CI: 95-100) in a patient proportion of 59% (53-66).
a Between patients proportions by paired Cochran test between the 4 strategies or paired McNemar test for pairwise comparison
Validation populations—With regard to the low constraint tests in validation population #1 (table 7), the most performant was again CirrhoMeterVIRUSV2G since providing the highest measured NPV for LEV—99% (97-100)—in the highest NPV patient proportion—60% (53-68)—(table 8).
)
100
)
0
)
19.4
b
)
)
)
0
)
)
)
0.8
)
)
5.6
)
)
)
3.6
)
98.9
74.2
)
25.8
0
)
3.2
b
)
0
)
92.4
)
0
)
0
)
a Measured predictive value in all patients
b Corresponds to missed LEV
c By paired Cochran test between the 3 tests
)
)
)
) a
0 ( )
)
)
0 ( )
a Corresponds to missed LEV
In validation populations #2 to #4 (2245 CLD patients: table 9), CirrhoMeterVIRUS2G was the test with the highest NPV patient proportion (298%) in non-cirrhotic patients across the 3 populations.
1.1
1.5
3.4
8.5
2.7
4.8
9.1
44.3
42.0
2.0
1.3
1.7
1.4
46.1
1.0
5.9
36.0
54.5
a No PPV zone (0% patients)
Derivation population—Among the five single test strategies (table 5), ECE was the most accurate test due to a significantly lower indeterminate patient proportion (29%, p<0.001) and the highest PPV patient proportion (11%). However, the measured PPV for LEV in this subgroup did not reach the targeted value: 80% in the largest population (table 6). Among the five combination strategies, two strategies ranked first for the two PPV criteria. Thus, ECE+(AST/ALT) score had the highest PPV patient proportion (10%) but was hampered by a suboptimal measured PPV—94% (77-100)—for LEV despite optimism bias. ECE+CirrhoMeterVIRUSS2G score reached the highest measured PPV for LEV at the expense of a lower PPV patient proportion than in other combinations (table 5). Measured PPV was 93% (75-100) in a patient proportion of 7% (3-10) in the largest population (table 6).
Validation populations—In validation population #1 (table 7), the three available blood tests showed similar results to derivation population (i.e. a high PPV was only observed in 1% of patients despite a 22% LEV prevalence). Thus, blood tests alone are not sufficiently predictive of the presence of LEV.
Among several clinically applicable strategies, the most accurate one seemed at this step to use first CirrhoMeterVIRUS2G mainly for LEV absence (i.e. the test with the highest NPV criteria), then to use the ECE+CirrhoMeterVIRUS2G score for LEV presence (i.e. a 93% PPV value in a substantial patient proportion). As there was a significant interaction (p<0.001) between CirrhoMeterVIRUS2G and ECE, we analyzed the plot of the two tests (
Robustness of test cut-off values for LEV prediction was evaluated in large unselected validation populations #2 to #4 (2245 CLD patients without decompensated cirrhosis) (table 9). Briefly, CirrhoMeterVIRUS2G robustness was validated, especially its estimated specificity was 100%, i.e. there was a priori no false positive LEV prediction in non-cirrhotic patients like in validation population #1.
All strategies were compared within the derivation population (table 5). Among the 7 clinically applicable strategies, the most accurate were the 3 combined sequential strategies since the two clinical descriptors (UGIE requirement and missed LEV) were better or equal than in the 4 single test strategies. Among these 3 combined sequential strategies, that including CirrhoMeterVIRUS2G offered the advantage of a higher rate of saved ECE (i.e. NPV patient proportion) or UGIE than the two others (p<0.001) with similar missed EV rate. Finally, the two strategies combining CirrhoMeterVIRUS2G and ECE (simultaneous or sequential) were directly compared (tables 6 and 10): the simultaneous combination resulted in significant lower missed LEV rate. However, the sequential strategy significantly reduced UGIE requirement (−7%) while saving 65% ECE.
2.8
61.1
33.3
5.6
72.0
−28.9
0
28.9
48.2
i
3.2
a The rates were calculated in the derivation population with maximum size (n = 211)
b Cut-offs of CM classes are those defined a priori for fibrosis stages in previous publication (Cales P et al, Journal of clinical gastroenterology 2014): cirrhosis is defined as possible (classes F3 ± 1, F3/4 and F4) or probable (classes F3/4 and F4) or very probable (classes F4). UGIE is performed only in the classes selected. The significance of gain could not be calculated since UGIE was performed in every patient
c By paired Cochran test between the 3 proportions. Pairwise comparisons by paired McNemar test: saved UGIE: all three pairs: p < 0.001; missed LEV: possible vs probable: p = 0.250, possible vs very probable: p = 0.001, probable vs very probable: p = 0.008
d Cut-offs of CirrhoMeterVIRUS2G classes and scores were defined a posteriori for LEV in the current derivation population. ECE is performed outside the PV zones of CM and UGIE is performed in the indeterminate zone
e By paired Cochran test between the 3 proportions. Pairwise comparisons by McNemar test for saved UGIE: CM vs CM/CM + ECE: p = 0.001, CM vs CM + ECE: p < 0.001, CM/CM + ECE vs CM + ECE: p = 0.038
f Comparison vs UGIE in histological cirrhosis by paired McNemar test. 95% CI cannot be calculated since this figure is obtained by a proportion difference
g By paired Cochran test between the 3 proportions. Pairwise comparisons by McNemar test: saved UGIE: all three pairs: p < 0.001; missed LEV: not calculable
h The simultaneous strategy based on CirrhoMeterVIRUS2G + ECE score was not evaluated since clinically unsuitable in a CLD population
i Estimated calculation by applying the rate of saved UGIE by CM + ECE score in the indeterminate CM zone from derivation population (36.6%)
k By paired McNemar test
We compared the direct LEV screening developed in the present study vs an indirect screening based on cirrhosis diagnosis (reference for UGIErequirement in the present study) or non-invasive fibrosis staging. Thus, we compared the three strategies including CirrhoMeterVIRUS2G (table 10). Briefly, in the validation CLD population, a strategy of performing endoscopy by the possible cirrhosis class of CirrhoMeterVIRUS2G would induce a significant UGIE overuse of 29% compared to conventional histological diagnosis of cirrhosis. Finally, at similar missed LEV rate, the saved UGIErate was much higher when CirrhoMeterVIRUS2G was targeted for LEV than for fibrosis, e.g. 18.4% (p=0.009) vs. 1.3% (p=1), respectively in derivation population.
Direct comparison of CirrhoMeterVIRUS2G, ECE and their combinations was performed in derivation population (
CirrhoMeterVIRUS2G was as accurate as ECE to predict LEV absence. However, ECE was significantly more accurate than CirrhoMeterVIRUS2G to predict LEV presence. Sequential combination significantly decreased the patient proportion with LEV presence from 12 to 8% compared to simultaneous combination but this was counterbalanced by an increase in measured PPV from 83% to 88%. The UGIE requirement by this sequential combination was significantly reduced when compared to CirrhoMeterVIRUS2G but not significantly different compared to ECE. The missed EV rate was significantly decreased by simultaneous combination compared to other strategies only in the derivation population. Finally, the sequential combination spared 48 to 72% of UGIE, whether UGIE would have been performed in all patients with cirrhosis, and spared 65% of ECE, whether ECE would have been performed in all CLD. The rate of correctly classified patients for LEV was, CirrhoMeter. 96.7% ((94.3-99.0), ECE: 90.0% (86.1-93.9), (CirrhoMeter+ECE) score: 98.6% (96.7-100), VariScreen algorithm: 96.2% (93.1-98.6), p<0.001 by paired Friedman test (pairwise comparison: ECE significantly lower than other tests and other test not significantly different between each other).
They were 21 patients (10.0%) misclassified for LEV by ECE; 5 were false positive LEV by ECE and 4 were rescued by VariScreen; there were 16 false negative LEV by ECE and 15 were rescued by VariScreen. Thus, 20 patients were rescued. However, there were 6 false negative LEV and 1 false positive LEV by VariScreen so that the net result was 20−7=13 corresponding to the 6.2% gain in accuracy by VariScreen compared to ECE. The 6 patients with missed LEV by VariScreen algorithm had blood markers significantly different (reflecting a better liver status) from other patients with LEV, e.g. serum albumin level (not included in CirrhoMeter): ruling out zone: 36.2±6.9 g/l, indeterminate zone: 34.1±5.8, ruling in zone: 25.9±5.1, p<0.001 by ANOVA.
The present study is the first one to compare ECE and fibrosis tests for the non-invasive diagnosis of LEV (Colli A et al, Cochrane Database Syst Rev 2014; 10:CD008760). In addition, studies of non-invasive diagnosis of LEV were performed in patients with cirrhosis selected by other means than non-invasive fibrosis tests. This casts some uncertainty about the cut-off exportability of non-invasive test cut-offs for LEV diagnosis in populations where cirrhosis will be diagnosed by the same non-invasive tests (for fibrosis staging there) among CLD. Therefore, we also evaluated together these tests in a population of CLD including non-cirrhotic patients with available UGIE which is a unique population. The only diagnostic combination algorithm published for high-risk EV was a sequential algorithm based on liver stiffness and concordant blood test in a first step followed by spleen stiffness in the intermediate zone; but the accuracy was only around 77% (Stefanescu H et al, Liver Int 2014).
Among fibrosis tests, blood tests appeared more interesting than VCTE for LEV diagnosis. This is due not only to a low maximum PPV for LEV but also to a lesser NPV patient proportion with VCTE (table 6). VCTE has been shown to well diagnose PHT level (Bureau C et al, Aliment Pharmacol Ther 2008; 27:1261-1268) but was limited and inferior to a single blood marker, like prothrombin index, for LEV diagnosis (Castera L et al, J Hepatol 2009; 50:59-68).
Among single tests, ECE was the most accurate non-invasive diagnosis for LEV providing the lowest rates of endoscopy requirement and missed LEV (table 5). In clinical practice, we have to choose a sequential diagnostic strategy based on a low constraint test to exclude LEV (expectedly in non-cirrhotic CLD) and on the most accurate test to diagnose LEV (expectedly used in cirrhosis). The most accurate sequential strategy was the VariScreen algorithm (
Patients between the two LEV cut-offs of ECE+CirrhoMeterVIRUS2G score are offered UGIE (
Finally, this study confirms that the non-invasive cirrhosis diagnosis has the potential to induce a roughly 30% endoscopy overuse. But applying cut-off of single fibrosis test specific for LEV is able to save one out 5 endoscopies compared to conventional strategy with cirrhosis diagnosis determined by liver biopsy (table 10).
The VariScreen Algorithm for LEV was not perfect with an indeterminate zone but it offered clinically relevant prediction with 88% PPV; moreover, the patients with false positive of VariScreen for LEV had small EV (table 11).
In addition, 96-99% NPV and 100% specificity were obtained in substantial patient proportions. The two latter figures were obtained with CirrhoMeterVIRUS2G which is the only available test specifically designed for cirrhosis diagnosis. It includes, hyaluronate which was the most accurate blood marker for LEV in the present study and elsewhere, and platelets and prothrombin index that are known markers for LEV.
The non-invasive diagnosis of LEV exhaustively applied to CLD is superior to the conventional attitude based on liver biopsy or clinics followed by endoscopy in all patients with cirrhosis in terms of saved endoscopy. Likewise, the use of specific cut off for LEV of a blood test spared endoscopy compared to the strategy using cut-off of this blood test for cirrhosis followed by endoscopy. This sparing effect can be significantly improved by ECE. Therefore, in the era of non-invasive testing, tests should be primarily focused on cirrhosis complications screening, like LEV, rather on cirrhosis diagnosis itself.
Screening for esophageal varices (EV) is recommended in cirrhosis. The Baveno6 recommendations allow ruling out EV if platelets >150 G/l and Fibroscan <20 kPa. However, primary prevention focuses on large EV (LEV) and it is unknown in which etiology this rule applies. Therefore, we evaluated this rule and tried to improve it with the aim of 100% predictive values (NPV, PPV).
287 patients with cirrhosis of various causes were prospectively included. Diagnostic tools were UGI endoscopy, 16 blood fibrosis tests, and Fibroscan. Patient characteristics were: men: 72%, age: 55+11 years, causes: alcohol: 64%, virus: 26%, NAFLD: 6%, others: 4%; EV: none: 56%, small: 27%, large: 17%.
Evaluation: NPV of Baveno6 rule was: EV: 87.1%, LEV: 100%. The spared endoscopy rate was only 16.4%. This rate was 38% with the best performing blood test (CirrhoMeter (CM), p<0.001 vs Baveno 6) for a missed LEV rate not significantly different (0%, 7%, respectively, p=0.157).
Improvement: A modified Baveno 6 rule (different cut-offs for platelets and Fibroscan) for EV had NPV100% in 18.2% of patients and even a PPV100% in 10.3% of patients. For LEV, there was a NPV100% in 37.0% of patients but no PPV100%. Finally, CM and Fibroscan combination had, respectively EV and LEV, NPV100% in 17.6% and 24.2% of patients and PPV100% in 6.7% and 3.0% of patients.
Discussion: The Baveno 6 rule has only a fair NPV for EV whereas it is very specific and poorly sensitive for LEV. New cut-offs provide NPV100% for LEV in more patients (37% vs 16%, p<0.001). By replacing platelets by a blood test, one can also get a 100% PPV. Thus, the best strategy is to use the modified Baveno 6 rule to rule out LEV and replace platelets by CM to rule in LEV. This algorithm has 100% accuracy with 0% missed LEV and 53.2% spared endoscopy. In practice, one measures platelets and stiffness in all patients; if the NPV100% cut-off is not reached, CM is performed; if the CM PPV100% cut-off is not reached, endoscopy is performed. The non-invasive strategy can be made in 1 or 2 steps knowing that the 2 non-invasive tests are already part of EASL and AASLD 2015 recommendations for fibrosis staging.
The Baveno 6 rule can be notably improved. With 2 simple non-invasive tests and without additional cost, it is possible not only to rule out but also to rule in LEV, which is original, with any missed LEV and half of endoscopies spared. These results have to be validated in another population.
The conventional management of patients with suspected liver cirrhosis suffers from several limitations. First, several surveys [5] have reported that LEV screening policies based on UGIE are not well applied, which is probably attributable to the aforementioned constraints encountered by physicians in real-life clinical practice. Second, classical liver biopsy cannot be easily repeated; this may allow asymptomatic cirrhosis to go undetected, only to be revealed later by the development of complications. Therefore, improving the non-invasive diagnosis of cirrhosis is indeed an attractive option, but this too has limits and implications. First, an earlier diagnosis introduces a risk of UGIE overuse (
The main objective of the present study was thus to develop a diagnostic strategy for LEV screening based on non-invasive and/or minimally-invasive tests. Toward this, ECE, liver elastography and fibrosis blood tests were tested, either alone or combined, in patients with cirrhosis. The secondary objective was to assess the exportability (i.e. generalizability) of the non-invasive LEV diagnostic strategy to the general CLD population, where non-invasive fibrosis test would be ultimately used.
The derivation population was extracted from a prospective study comparing ECE and UGIE for the diagnosis of LEV (large esophageal varices) in patients with cirrhosis of various etiologies recruited from April 2010 to March 2013 [8]. The 287 patients in whom both ECE and UGIE were performed were included. Diagnostic algorithms were developed in this derivation population of patients with cirrhosis.
The validation population included 165 patients with CLD attributed to viral infection or alcohol use, with or without cirrhosis, who had all undergone UGIE [9, 10]. This was a prospective study where UGIE was indicated to evaluate PHT signs. Blood tests and liver biopsy were available for all of the patients and all fibrosis stages were represented. However, these patients did not undergo ECE. Thus, this population was used to validate only the non-invasive strategy.
Endoscopic procedures are detailed elsewhere [8, 11]. In both populations, EV size was classified into three grades: small, medium or large [10]. The two last grades were grouped as LEV. Sixteen blood tests were calculated (details in supplemental material). Among them, CirrhoMeterV2G, called CirrhoMeter hereafter, offered the highest performance. Vibration-controlled transient elastography (VCTE) (Fibroscan, Echosens, Paris, France) was performed according to the manufacturer's recommendations [12] by operators blinded to the other results.
In this kind of study, only direct costs can be calculated. The costs of tests were those of the French list of care costs: CirrhoMeter: €29, UGIE: €114 (applying the 38% rate of general anesthesia recorded in the pivotal study), ECE: €612, liver biopsy: €1050 (including day hospitalization). The direct costs were calculated in the validation population, which was the only population where both non-invasive fibrosis and LEV tests were evaluated in the setting of clinical care.
Ten diagnostic strategies were evaluated (see Table 12): five comprised a single diagnostic test and five a combination of several tests. These combinations were either simultaneous or sequential. When we were developing strategies, the priority objective was a missed LEV rate ≤5%.
59.5
29.1
a
11.4
10.0
96.8
83.3
b
52.5
0
c
20.0
93.0
100
78.5
11.4
10.1
26.7
95.8
100
93.3
58.9
29.7
11.4
97.8
aPatients diagnosed with small EV by ECE
b It was not possible to reach the objective (≥90%) as this is a semi-quantitative variable
c Maximum PPV was <40%
d This figure is different from ECE alone because three patients with high LEV PPV with ECE had high LEV NPV with VCTE
e This figure is different from (ECE + (AST/ALT)) score alone (simultaneous combination) because two patients with high LEV PPV with (ECE + (AST/ALT)) score had high LEV NPV with VCTE
f CirrhoMeter + (CirrhoMeter + ECE) score
g By paired Cochran test for patient proportions
Clinically applicable sequential strategies were selected. A clinically applicable strategy was defined as one including obligatorily a low constraint test (e.g. blood test) to rule out LEV (usually in asymptomatic patients) and possibly a high constraint test (e.g. UGIE) to rule in LEV (usually in the most severe patient cases).
The most predictive of several clinically applicable strategies (details in Table 12) was to use CirrhoMeter first, mainly to rule out LEV (i.e. the test with the highest NPV criteria), then the combination of ECE and CirrhoMeter into a score to rule in LEV (positive predictive value (PPV)=93%). As there was a significant interaction (p<0.001) between these two tests, we analyzed their scatter plots (
In practical terms, the VariScreen algorithm is as follows: CirrhoMeter is performed in all patients. Those with CirrhoMeter ≤0.21 are followed-up with yearly CirrhoMeter testing. Those with CirrhoMeter >0.9994 are offered primary prophylaxis. Patients between these two CirrhoMeter cut-offs are offered ECE. Then, the (ECE+CirrhoMeter) score is calculated by computer. Patients with (ECE+CirrhoMeter) scores <0.1114 are followed-up with yearly CirrhoMeter testing. Those with (ECE+CirrhoMeter) scores >0.55 are offered primary prophylaxis. Finally, patients between the two score cut-offs are offered UGIE, as they run a 23% probability of having LEV.
As VariScreen includes ECE, this is a partially minimally-invasive strategy. Therefore, an entirely non-invasive strategy was also developed.
Missed LEV—This was the proportion of patients with undetected LEV in the patient subgroup with LEV.
UGIE requirement—This was the proportion of patients in the indeterminate zone of the non-invasive tests, i.e. between their negative predictive value (NPV) and PPV cut-offs for LEV.
Spared UGIE—Patients with cirrhosis were used as the reference group to calculate the rate of patients that the algorithm would spare from UGIE, as this latter is classically performed in these patients. This comprised the entire derivation population and patients with Metavir F4 stage by liver biopsy or with cirrhosis diagnosed by CirrhoMeter in the validation population. The spared UGIE rate corresponds to the difference between the cirrhosis group and the LEV target group where UGIE was indicated by non-invasive tests. Thus, non-invasive tests might be used with cut-offs for cirrhosis diagnosis or LEV diagnosis.
For LEV diagnosis, we initially determined the two cut-offs of a test value to reach a NPV≥295% and a PPV≥290%. Consequently, these two cut-offs determined three diagnostic zones: LEV ruled out (≤NPV cut-off), indeterminate, and LEV ruled in (≥PPV cut-off). In the final diagnostic algorithm, the cut-offs of constitutive tests were adjusted to minimize the missed LEV rate (priority clinical objective) if necessary.
Quantitative variables were expressed as mean±standard deviation. The discriminative ability of each test was expressed as the area under the receiver operating characteristic (AUROC) curve and compared by the Delong test. Data were reported according to STARD [13] and Liver FibroSTARD [14] statements, and analyzed on an intention-to-diagnose basis. Scores including independent LEV predictors were determined by binary logistic regression. The main statistical analyses were performed under the control of professional statisticians (SB, GH) using SPSS version 18.0 (IBM, Armonk, N.Y., USA) and SAS version 9.3 (SAS Institute Inc., Cary, N.C., USA).
The characteristics of the populations are provided in Table 2 hereinabove. In the validation CLD population, LEV were only observed in patients with cirrhosis confirmed by liver biopsy and in those with probable cirrhosis (estimated Metavir classes F3/4 and F4) according to CirrhoMeter.
AUROCs for LEV are detailed in Table 4 hereinabove. Briefly, in the derivation population, the highest AUROCs (≥0.91) were obtained with two scores combining ECE and blood markers. The AUROCs of these scores were significantly higher than those of fibrosis tests (p<0.02), whereas the AUROCs between fibrosis tests were not significantly different (details not shown). In the validation population, the AUROC of Metavir F stages by liver biopsy (0.819) was significantly lower (p<0.01) than that of the most accurate blood fibrosis tests (e.g. 0.911 for CirrhoMeter).
CirrhoMeter was the best performing low constraint test, providing the largest ruled out zone and the highest measured NPV for LEV (Table 13).
ECE was the best of the five single test strategies (details in the supplemental material), providing a significantly lower proportion of indeterminate patients and the largest ruled in zone. However, its PPV for LEV was 80% (Table 13), falling short of the targeted 90% value. Among the five combination strategies, the (ECE+CirrhoMeter) score provided the highest measured PPV for LEV (Table 13).
a With positive predictive value for LEV
b CirrhoMeter + (CirrhoMeter + ECE) score
c Patients proportions: between the four strategies by paired Cochran test and for pairwise comparison by paired McNemar test
Table 14 shows that CirrhoMeter targeted for cirrhosis had to be used with its three classes including F4 to miss <5% LEV. Spared UGIE was then 15.6%. However, CirrhoMeter and the CirrhoMeter+FibroMeter algorithm targeted for LEV significantly increased (p<0.001) spared UGIE to 36.0 and 43.1%, respectively, the latter figure being significantly higher than the former (p<0.001). In other words, targeting CirrhoMeter to LEV reduced UGIE by 14.4% (p<0.001) compared to targeting it for cirrhosis.
a CirrhoMeter targeted for LEV
b The rates were calculated in the derivation population with maximum size (n = 211)
c CirrhoMeter fibrosis classification includes 6 classes, 3 of which include F4: F3 ± 1 + F3/4 + F4
d PPV is artificially at 100% due to cirrhosis population selection
eCorresponds to the indeterminate zone for LEV; there was no PPV zone for LEV in the validation population
Comparisons for the VariScreen algorithm and its constitutive tests are presented in Table 15. Briefly, the missed LEV rates were not significantly different between the tests. ECE accuracy was significantly lower than in other tests. Considering spared UGIE rates, all tests were significantly different except VariScreen and ECE. Thus, there was a progressive increase in spared UGIE, CirrhoMeter targeted for cirrhosis: 15.6%, CirrhoMeter targeted for LEV: 36.0% (p<0.001 vs previous), CirrhoMeter+FibroMeter algorithm: 43.1% (p<0.001 vs previous), ECE and VariScreen: around 70% (p<0.001 vs previous).
a Correctly classified patients for LEV
b CirrhoMeter with cut-off targeted for cirrhosis
c CirrhoMeter with adjusted cut-off targeted for LEV
d Paired Cochran test for global comparison and paired McNemar test for pair comparisons
VariScreen accuracy was not dependent on Child-Pugh classes: A: 97.7%, B: 94.4%%, C: 97.3%, p=0.587).
The distribution of small EV and gastric varices as a function of the VariScreen algorithm is depicted in Table 16.
1
3
Misclassified patients—Twenty-one patients (10.0%) were misclassified for LEV by ECE, including 5 false positives of which 2 were rescued by VariScreen (
Six patients were misclassified for LEV by VariScreen, specifically 4 false positives and 2 false negatives. The 2 false negative patients had blood markers significantly different (reflecting a better liver status) from other patients with LEV, e.g. median serum albumin levels (g/l) in patients with LEV: ruled out zone (i.e. the 2 missed LEV): 41.5; indeterminate zone: 31.0; ruled in zone: 27.0; p=0.040 by Kruskal-Wallis test. Among the 4 false positive cases, 3 had small EV, explaining the VariScreen PPV for EV of 97%.
Comparison with recommendation—The Baveno VI rule had high NPV: 86.2% for EV and 100% for LEV, but the spared UGIE rate was only 18.4% vs 70.3% (p<0.001) with VariScreen or 38.0% with CirrhoMeter targeted for LEV (p<0.001) while missed LEV rates were not significantly different (details in Table 17).
0
70.3
a Cut-offs of CirrhoMeter and scores were defined a posteriori for LEV in the derivation population
b Cut-offs of CirrhoMeter used alone, either unadjusted or adjusted (as used in VariScreen)
c Spared UGIE when VCTE < 20 kPa and platelets > 150 G/l
d CirrhoMeter + (CirrhoMeter + ECE) score
e By paired Cochran test between the four proportions. Pairwise comparisons by paired Wilcoxon test
The CirrhoMeter and the CirrhoMeter+FibroMeter algorithm targeted for LEV did not significantly reduce UGIE compared to cirrhosis diagnosis by liver biopsy but they did compared to CirrhoMeter targeted for cirrhosis, e.g. 21.9% (p<0.001) for CirrhoMeter+FibroMeter algorithm (Table 14). Importantly, the missed LEV rate was 0%.
The strategies with minimal missed LEV rates (0.3%) are analyzed in terms of costs. The most expensive strategy was the classical strategy based on initial cirrhosis diagnosis by liver biopsy (Table 18). The least expensive strategy was that based on CirrhoMeter or CirrhoMeter+ FibroMeter targeted for LEV. The addition of ECE multiplied the cost of the latter by 4.2 (or 3.1 vs CirrhoMeter targeted for cirrhosis) but VariScreen was 3.5 times less expensive than the classical strategy based on fibrosis staging by liver biopsy.
a The reference population is cirrhosis unless otherwise stated
b The reference population is cirrhosis diagnosed by CirrhoMeter
c No additional cost for FibroMeter
d Calculation estimated by applying the rate of spared UGIE by (CirrhoMeter + ECE) score in the indeterminate CirrhoMeter zone of the derivation population (48.1%) assuming a robust ECE performance and using the cut-offs of the VariScreen algorithm
e Calculation assuming that missed LEV are attributable to CirrhoMeter
The objective is to obtain a non-invasive diagnosis of large esophageal varices (LEV) with the following rules for statistical algorithms:
CirrhoMeterV2G called CirrhoMeter (CM) thereafter and expressed as a score from 0 to 1.
FibroMeterV2G called FibroMeter (FM) thereafter and expressed as a score from 0 to 1. Platelet (Pl) count expressed in G/l.
Fibroscan (FS) called vibration controlled transient elastography (VCTE) thereafter and expressed in kPa
Cirrhotic patients from the VO-VCO studies (Sacher-Huvelin Endoscopy 2015) (see description hereinabove in Examples 1 and 3):
They are based on single negative predictive value (NPV) zone and positive predictive value (PPV) zone according to classical statistical rules.
This FibroMeter+CirrhoMeter algorithm for large esophageal varices is described in Example 3 (see
The LEV rule out (NPV) zone is defined by the following cut-offs: CirrhoMeter <0.21 or FibroMeter <0.78.
The LEV rule in (PPV) zone is defined by the following cut-offs: CirrhoMeter >0.998 and FibroMeter >0.9993.
With this second FibroMeter+CirrhoMeter algorithm with different cut-off values, there is no missed LEV and less false positives (only one) compared to CMFM#1.
The LEV rule out zone is defined by the following cut-offs: CirrhoMeter <0.042 or FibroMeter <0.51
The LEV rule in zone is defined by the following cut-offs: CirrhoMeter >0.99945
This is a Platelets+VCTE (also known as Fibroscan™) algorithm for large esophageal varices.
The LEV rule out zone is defined by the following cut-offs: platelets >110 G/l and VCTE<26.5 kPa.
The LEV rule in zone is defined by the following cut-offs: platelets <45 G/l and VCTE>32 kPa.
NB: PPV is 100% in 0 patients, i.e. no PPV zone.
This is another Platelets+VCTE (also known as Fibroscan™) algorithm for large esophageal varices.
New rule for LEV rule in zone: presence with a minimum of false positives (only one in fact).
The LEV rule out zone is defined by the following cut-offs: platelets >110 G/l and VCTE<26.5 kPa.
The LEV rule in zone is defined by the following cut-offs: platelets <65 G/l and VCTE>32 kPa.
PPV is 83.3% among 6 patients in a sample size of 165 patients
PPV is 85.7% among 7 patients in a sample size of 221 patients
This CirrhoMeter+Fibrocan™ (also known as VCTE) algorithm is described hereinabove in Example 2.
The LEV rule out zone is defined by the following cut-offs: CirrhoMeter <0.6 and VCTE<14 kPa.
The LEV rule in zone is defined by the following cut-offs: CirrhoMeter >0.99891 and VCTE>55 kPa.
They are based on multiple negative predictive value (NPV) zones and positive predictive value (PPV) zones according to new statistical rules as described in Example 5 below.
This is a Platelets+CirrhoMeter+VCTE (also known as Fibroscan™) algorithm for large esophageal varices.
LEV rule in zone: CirrhoMeter >0.99891 and VCTE>55 kPa.
This is a Platelets+FibroMeter+CirrhoMeter+VCTE (also known as Fibroscan™) algorithm for large esophageal varices.
Baseline algorithm: PlCMFS#1
0.009
a five false positive
b one false positive
aone false positive
PlFS#1 corresponds to tests used in Baveno 6 rule for EV NPV (De Franchis J Hepatol 2015) but cut-offs are here specific to LEV. One can add a NPV zone (PlFS#1b).
One can use only blood tests (FM/CM or CMFM#1) with accuracy slightly superior to modified Baveno 6 rule by accepting a small proportion of missed LEV not significantly from 0% of Baveno 6 rule. Otherwise, by targeting 0% missed LEV (CMFM#1b), the spared UGIE rate is significantly decreased.
The combination of CM and VCTE allows no missed LEV but with a spared UGIE rate significantly decreased compared to the modified Baveno 6 rule.
The combination of the modified Baveno 6 rule (PlFS#1) for LEV out zone (with an additional zone) to the CM and VCTE combination for LEV in zone (PlCMFS#1) associates respective advantages with significantly increased spared UGIE rate compared to each constitutive algorithm. Additional zones for LEV in or out zone (PlFMCMFS#1) increased this rate at the expense of possible overfitting (optimism bias).
This is a Platelets+VCTE (also known as Fibroscan) algorithm for esophageal varices. The EV rule out zone is defined by the following cut-offs: platelets >87 G/l and VCTE<11.9 kPa.
The EV rule in zone is defined by the following cut-offs: platelets <93 G/l and VCTE>30 kPa.
The cut-offs for NPV zone are improved compared to original Baveno 6 rule; PPV zone is also an improvement.
With 0=NPV zone (LEV rule out zone), 1=grey zone, 2=PPV zone (LEV rule in zone).
compute CMFM#1=1.
compute CMFM#1=0.
else if (FibroMeter <0.78).
compute CMFM#1=0.
else if (CirrhoMeter >0.998) and (FibroMeter >0.9993).
compute CMFM#1=2.
end if.
execute.
compute CMFM#1b=1.
compute CMFM#1b=0.
else if (FM2G<0.51).
compute CMFM#1b=0.
else if (CM2G>0.99945).
compute CMFM#1b=2.
end if.
execute.
compute PlFS#1=1.
do if (platelets >110) and (VCTE<26.5).
compute PlFS#1=0.
else if (platelets <45) and (VCTE>32).
compute PlFS#1=2.
end if.
execute.
compute PlFS#1=1.
do if (platelets >110) and (VCTE<26.5).
compute PlFS#1=0.
else if (platelets <65) and (VCTE>32).
compute PlFS#1=2.
end if.
execute.
compute PlFS#2=1.
do if (platelets >87) and (VCTE<11.9).
compute PlFS#2=0.
else if (platelets <93) and (VCTE>30).
compute PlFS#2=2.
end if.
execute.
compute CMFS#1=1.
compute CMFS#1=0.
else if (CirrhoMeter >0.99891) and (VCTE>55).
compute CMFS#1=2.
end if.
execute.
compute PlCMFS#1=1.
do if (platelets >110) and (VCTE<26.5).
compute PlCMFS#1=0.
else if (CirrhoMeter <0.334) and (VCTE<35).
compute PlCMFS#1=0.
else if (CirrhoMeter >0.99891) and (VCTE>55).
compute PlCMFS#1=2.
end if.
execute.
compute PlFMCMFS#1=1.
do if (platelets >110) and (VCTE<26.5).
compute PlFMCMFS#1=0.
else if (CirrhoMeter <0.334) and (VCTE<35).
compute PlFMCMFS#1=0.
else if (CirrhoMeter <0.04).
compute PlFMCMFS#1=0.
else if (FibroMeter <0.05).
compute PlFMCMFS#1=0.
else if (FibroMeter <0.895) and (VCTE<33).
compute PlFMCMFS#1=0.
else if (VCTE<9.1).
compute PlFMCMFS#1=0.
else if (CirrhoMeter >0.99891) and (VCTE>55).
compute PlFMCMFS#1=2.
else if (FibroMeter >0.9994) and (VCTE>60).
compute PlFMCMFS#1=2.
end if.
execute.
This example describes a method to determine a diagnostic algorithm based on multiple diagnostic tests by using their respective predictive values.
The data supporting this description are drawn from the diagnosis of esophageal varices in cirrhosis.
The objective of the method of multiple zones of predictive values is to increase the predictive value of a diagnostic algorithm by combining the predictive values of at least 3 diagnostic tests (or markers).
Frequently, diagnostic tests cannot be sorted as binary with a yes/no result and a single cut-off.
The main solution is to consider predictive values and to accept a maximal error risk, e.g. 5% even 0%. Thus, one has to calculate two cut-offs: one for negative predictive value (NPV) and one for positive predictive value (PPV).
The cut-offs for a single diagnostic test are calculated as shown in
Then, the predictive zones of a single diagnostic test are obtained as shown in
It can be more accurate to define predictive zones using two diagnostic tests as shown in
It is more difficult to calculate predictive zones by using more than two tests. A new method to solve this difficulty is described below.
The principle is the following.
The NPV and PPV zones are determined according to classical rules described in the hereinabove background paragraph.
Thus, these zones have to be plausible, i.e. zones have to be located in the expected values of the diagnostic test for the corresponding predictive value, e.g. platelet count in the highest range to rule out (NPV zone) large esophageal varices.
The predictive zone obtained with two tests can be calculated in several ways:
This method bears the risk of maximizing the optimism bias; therefore, the following precautions have to be taken:
| Number | Date | Country | Kind |
|---|---|---|---|
| 15162685.0 | Apr 2015 | EP | regional |
| 16163029.8 | Mar 2016 | EP | regional |
| Filing Document | Filing Date | Country | Kind |
|---|---|---|---|
| PCT/EP2016/057653 | 4/7/2016 | WO | 00 |
