Vibration Controlled Transient Elastography (Fibroscan®) in sickle cell liver disease - could we strike while the liver is hard?

Summary

Vibration controlled transient elastography (VCTE) is validated for the evaluation of hepatic fibrosis in different liver diseases. Sickle cell liver disease (SCLD) results from a cumulative hepatic injury and its lifelong and progressive nature raises the need for a non-invasive tool for fibrosis evaluation.

Fifty patients, aged between 23 and 59 years with sickle cell disease and suspected SCLD underwent a VCTE followed by a liver biopsy. Biopsies were evaluated for various scores of liver disease that were then correlated to VCTE score. 90% of our patients had an Ishak Fibrosis (IF) score between 0–2 (Group A-minimal to no fibrosis) and 10% of the patients had IF score between 3–6 (Group B-advanced fibrosis). The median Transient Elastography (TE) for patients in Groups A and B was 4·8 kilopascals (kPa) and 17·6 kPa, respectively. A positive correlation was shown between TE and IF score, R = 0·0·68 (P = <0·0001); a positive correlation was also shown with Histology Activity Index fibrosis score, R = 0·64 (P = <0·0001). This study emphasises the need for further studies of non-invasive tools and their utility in liver fibrosis evaluation of patients with SCLD.

INTRODUCTION

Sickle cell liver disease (SCLD), otherwise known as sickle cell hepatopathy, is a term describing the hepatic manifestations of sickle cell disease (SCD) (Diggs, 1992), in which it is associated with significant morbidity and mortality in affected patients (Feld et al., 2015). SCLD, first described in 1953, presents a phenotype that extends from mild inflammation to cirrhosis (Green et al, 1953; Bogoch et al, 1955).

Liver injury in SCLD is a result of multiple precipitating factors, such as recurrent ischaemic episodes, prolonged inflammatory exposure secondary to intra-sinusoidal erythrocytes sickling, persistent hepatocyte regeneration and iron overload driven by haemolysis and frequent blood transfusions (Shah et al, 2017). Moreover, the frequent need for blood transfusions, a common feature in SCD, exposes patients to blood-borne hepatotropic viral infections, such as hepatitis B and C (Issa, 2010; Bakarey et al, 2018).

Currently, liver biopsy is still considered the gold standard for determining hepatic fibrosis and is the only diagnostic modality and follow-up tool for SCLD.

In addition to its invasive nature, Zakaria et al. (2003) have shown that liver biopsy in SCLD is high risk when performed in close proximity to a pain crisis. This limits its utility as an accessible tool for diagnosis and follow-up of SCLD.

The development of non-invasive screening and follow-up tools for the evaluation of SCLD progression would therefore be highly desirable. Such a tool could potentially allow early intervention and management modification in the treatment of the primary disease.

Vibration controlled transient elastography (VCTE) is a non-invasive modality validated for the evaluation of tissue fibrosis.

In many entities of liver diseases, the use of transient elastography (TE) has reduced the need for liver biopsy in the estimation of fibrosis (Njei et al., 2016; Singh et al., 2017; Siddiqui et al., 2019). This technique is relatively simple and can be performed during clinic visits.

TE uses tissue propagation of shear wave vibration as an assessment of tissue stiffness. It is expected that, as liver tissue becomes more fibrotic, its stiffness increases, hence the shear wave will propagate more rapidly within the tissue, causing the TE score to increase. TE was previously shown to increase during acute sickling episodes and correlate with increased markers of hepatocellular injury in patients with SCD (Koh et al., 2013). This, however, was not paired with histological scores, and its role as an evaluation tool in the presence and extent of liver involvement in SCD is yet to be evaluated (Koh et al., 2013; Drasar et al., 2017).

This study aimed to evaluate the correlation of VCTE to histological scores of inflammation, fibrosis and iron load in liver biopsy tissue in close proximity to the elastography measurements.

METHODS

Study population

This is a retrospective study. Patients followed by the Sickle Cell Branch with clinical, laboratory or imaging findings suggestive of liver involvement in the primary haematological disease were referred for further evaluation to the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK), Liver Diseases Branch. These patients consented to the Liver Diseases Branch, Liver Disease Natural History Protocol (NCT00001971). Medical records of 50 patients were obtained and reviewed between 2004 and 2018.

Data collection & histological evaluation

Fifty patients aged between 23 and 59 years were included in the study. All patients had a pre-liver biopsy VCTE evaluation. They subsequently underwent an elective diagnostic liver biopsy within 3 months irrespective of their VCTE results. VCTE, measured via Fibroscan® (Echosens, Paris, France), was evaluated in patients after informed consent. Physicians specifically trained in the technique conducted the elastography measurements. A successful measurement was validated using the following three criteria recommended by the manufacturer: (i) at least 10 determinations; (ii) an interquartile range (reflecting the variability of measurements) of less than 30% of the median Liver Stiffness (LS) measurement value; (iii) a success rate (ratio of valid shots to the total number of shots) of at least 60%. The median results of LS were expressed in kilopascals (kPa). All liver biopsies (percutaneous or transjugular) were obtained and evaluated within three months of their VCTE results, by a single pathologist who was blinded to the patients’ VCTE results. The biopsy specimens were stained and scored for Ishak fibrosis (IF) score, Histology Activity Index (HAI) fibrosis and HAI inflammation scores, and iron score (Rowe et al., 1977; Knodell et al., 1981; Ishak et al., 1995). Dry liver sample was sent for iron quantification and hepatic iron index was calculated. The hepatic iron index was calculated from the hepatic iron concentration by dividing the quantitative iron (μmol/g dry liver weight) with the age of the patient (Kowdley et al., 1997). A histopathological picture of SCD taken by hepatopathologist, DEK, can be seen in Fig 1. Laboratory evaluations, such as alkaline phosphatase (ALP), alanine aminotransferase (ALT), aspartate aminotransferase (AST), serum iron and serum ferritin levels and platelet counts were obtained simultaneously at the time of liver biopsy.

Fig 1.

Liver tissue of a patient with sickle cell liver injury, showing marked hepatic haemosiderosis, mild inflammation and bridging fibrosis (1A: Iron stain, 100×; 1B: Haematoxylin and eosin stain, 100×; 1C: Masson stain, 100×).

Statistical analysis

Continuous variables were reported as medians and interquartile ranges (IQR). Binary variables were reported as frequencies. Due to uneven distribution of patients along the fibrotic scale (based on IF score on liver biopsy) and only a small number of patients with advanced fibrosis, all subjects were categorized into two groups: Group A and Group B. Group A had patients with IF score 0–2 (minimal to no fibrosis) and Group B had patients with IF score 3–6 (advanced fibrosis). Non-parametric Mann-Whitney tests were used to evaluated differences between Group A and Group B. To assess for differences in TE across IF in Group A, we conducted a Kruskal-Wallis test followed by post-hoc analysis using the Tukey-Kramer test for multiple comparisons. All analyses were performed using GraphPad Prism® v.7a (GraphPad Software, San Diego, CA, USA) and SPSS v.20 (IBM, Armonk, NY, USA). Correlations were then measured for the entire cohort between different parameters by the Spearman’s rank-order correlation. P-value <0·05 was used to define statistical significance.

IRB approval

The Institutional Review Board of the National Institute of Diabetes and Digestive and Kidney Diseases (NIDDK) approved the clinical research study prior to data collection.

RESULTS

Patient demographics

A total of 50 patients (21 males) were evaluated; 90% of the patients had IF score between 0–2 (Group A-minimal to no fibrosis) and 10% of the patients had IF score between 3–6 (Group B-advanced fibrosis). Out of all these patients, 44 were African Americans, 4 were multiracial, 1 was Caucasian and 1 was Asian. Median age was 36 years (Interquartile range: 29, 42 years). Forty-six patients had homozygous SCD (HbSS), 2 had Haemoglobin SC disease (HbSC) and 2 had Sickle beta-thalassaemia (HbSβ) (Table I).

Table I.

Demographic and serological characteristics of all patients, and patients according to IF score (Group A or B), represented as a median value with interquartile ranges.

Features	All subjects n = 50	Grouped by Ishak fibrosis score
		Group A (IF 0–2)n = 45	Group B (IF 3–6)n = 5	P-value†
Patients (n)	50	45	5
Gender (Male)	21	21	0
Age at biopsy, years	36 (28, 42)	36 (23, 59)	50 (33, 50)	0·003
Race (AA/C/A/M)	44/1/1/4	42/1/1/1	2/0/0/3
Genotype SS/SC/Sβ	46/2/2	42/2/1	4/0/1
Laboratory tests
Serum iron (μmol/l)	21·2 (14·5, 31·7)	21·1 (14·4, 31·3)	28·8 (12·1, 42·9)	0·9
Ferritin (μg/l)	1076 (356·3, 3197)	1134 (426, 2716)	380 (106, 7505)	0·87
ALT (u/l)	27·5 (21, 45·5)	27 (21, 40·5)	52(25·5, 57)	0·17
AST (u/l)	37(28·75, 56·25)	36 (27·5, 54·5)	61 (41, 88)	0·04
Platelet count (×109/l)	282 (219·8, 374·5)	286 (240, 403·5)	206 (170, 290·5)	0·08
Albumin (g/l)	39·5 (38, 42)	40 (39, 42)	38 (37, 41)	0·37
International Normalised Ratio	1·15 (1·07, 1·22)	1·15 (1·07, 1·23)	1·16 (1·1, 1·2)	0·7

A, Asian; AA, African American; ALT, Alanine Aminotransferase; AST, Aspartate Aminotransferase; C, Caucasian; IF, Ishak fibrosis; M, multiracial.

^†(Mann-Whitney test for Unpaired t-test).

Comparisons between patients with minimal & advanced fibrosis on biopsy

The median VCTE for all patients was 5·25 kPa, with the medians of 4·8 kPa and 17·6 kPa for patients in Group A and B, respectively. Within Group A, there was a trend in TE recordings: for Ishak 0, Median TE was 4·6 (IQR 3·5, 6·1) kPa; for Ishak 1, Median TE was 9·6 (IQR 4·4, 12) kPa; for Ishak 2, Median TE was 14 (IQR11·2, 19·25) kPa. One way analysis of variance followed by post-hoc analysis using the Tukey-Kramer test for multiple comparisons showed significant difference in TE between patients with IF0 to those with IF2 (P = 0·0012).

The demographic and serological characteristics are shown in I. Fifty of the biopsies were scored: 34 with IF0, 6 with IF1, 5 with IF2, 2 with IF3, 0 with IF4, 0 with IF5 and 3 with IF6. The median HAI fibrosis scores in Group A and B were 0 and 4 respectively whereas the median HAI inflammation scores were 1 and 4 in the two groups. Median Iron score in Group A was 3 and Group B was 2. Median Iron content in liver sample of all patients was 4219 μg/g (Table II). All of these values were steady state values obtained within three months of patients’ respective TE readings.

Table II.

Surrogates of fibrosis and liver biopsy characteristics of all patients, s and patients according to IF score (Group A or B),represented as a median value with interquartile ranges.

	All subjects n = 50	Grouped by Ishak fibrosis score
		Group A (IF 0–2)n = 45	Group B (IF 3–6)n = 5	P-value*
Surrogate for fibrosis
APRI	0·41(0·27, 0·62)	0·36 (0·25, 0·54)	0·85(0·61, 1·34)	0·008
Fib-4	0·87 (0·57, 1·37)	0·84 (0·56, 1·23)	1·72 (1·54, 2·81)	0·002
Fibroscan® (kPa)	5·25 (3·7, 10·8)	4·8 (3·65, 8·4)	17·6 (11·4, 30·35)	0·0004
Liver biopsy
HAI fibrosis score	2·4 (0·72, 7·88)	0 (0,0)	4 (3,4)
HAI inflammation score (0–18)	0 (0, 1)	1 (1,3)	4 (3·5,7·5)
Iron score	1 (0, 3·25)	3 (1,4)	2 (1,4)
Iron, liver (mg/g dry weight)	4219 (1370, 144774)	5072 (1483, 14474)	1051 (380·8, 1310)	0·31
Iron index (mmol/g/year)	2·4 (0·73, 7·9)	2·65 (0·93, 7·88)	0·5 (0·15, 7·83)	0·16

APRI; aspartate aminotransferase/platelet ratio index; HAI: histology activity index; IF: Ishak fibrosis

VCTE in SCLD

A positive and significant correlation was shown between TE and IF (R = 0·68, P = <0·0001), a positive and significant correlation was also found with HAI fibrosis score (R = 0·64, P = <0·0001) and HAI inflammatory score (R = 0·44, P = 0·001) (Fig 2). There was no significant correlation between TE and Liver biopsy Iron score, or between Iron quantification in the liver and Iron Index. Weak significant correlation was seen between TE and serum iron (R = 0·3, P = 0·028) as well as with serum ferritin (R = 0·3, P = 0·03) There was no significant correlation between TE and laboratory values of liver-related enzymes or platelets (III).

Fig 2.

Vibration controlled transient elastography (Fibroscan®) score compared to biopsy fibrosis evaluated by Ishak fibrosis score.

We calculated the following parameters: sensitivity, specificity, positive predictive value (PPV) and negative predictive value (NPV) for VCTE as the investigation tested for advanced fibrosis screening and liver biopsy IF score as gold standard for measuring liver fibrosis. IF score of 0–2 was considered as absence of fibrosis and IF score of 3–6 was considered as presence of fibrosis. Two separate cut-offs values of TE score of 7·5 kPa and 8·5 for diagnosis of hepatic fibrosis were used. A TE cut-off of 7·5 kPa, had a sensitivity of 1, specificity of 0·66, PPV of 0·25 and NPV of 1. On increasing the TE cut-off to 8·5 kPa, although the sensitivity and NPV both remained 1, the specificity and PPV increased to 0·75 and 0·31 respectively (Fig 3, Table IV).

Fig 3.

Statistical validity of vibration controlled transient elastography as a measure of hepatic fibrosis with two separate cut-offs for transient elastography [Gold standard test: Liver biopsy Ishak fibrosis (IF) score; IF score 0–2: negative fibrosis; IF score 3–6: positive fibrosis]. NPV, negative predictive value; PPV, positive predictive value; TE, transient elastography.

Table IV.

Measures of statistical validity to evaluate Fibroscan® as a measure of hepatic fibrosis with two separate cut-offs for transient elastography [Gold standard test: Liver biopsy Ishak fibrosis (IF) score; IF score 0–2: negative fibrosis; IF score 3–6: positive fibrosis]

	TE cut-off 7·5	TE cut-off 8·5
True positive, false negative	5, 0	5, 0
True negative, false positive	30, 20	34, 11
Sensitivity/ Specificity	1·0, 0·66	1·0, 0·75
PPV/NPV	0·25, 1·0	0·31, 1·0

NPV, negative predictive value; PPV, positive predictive value; TE, transient elastography.

DISCUSSION

Sickle Cell disease is a haematological disorder of haemoglobin polymerization, which was first described in 1910 by J. B. Herrick (Herrick, 1910; Rees et al, 2010). There are 100,000 people in the United States with SCD, the prevalence being 1 out of every 365 African Americans and 1 out of every 16 300 Hispanic-Americans (Centers for Disease Control & Prevention, 2019). The disease is a spectrum of various organ system complications. One of the common manifestations of SCD is an affected hepatobiliary system (Banerjee et al, 2001). Described for the first time in the 1950s as a cause of cirrhosis, SCLD (or sickle cell hepatopathy) has been better understood only in the past several decades (Green et al, 1953; Bogoch et al, 1955). The injury is multifactorial, attributed to hypoxic insult from the sickled RBCs in the liver sinusoids leading to hepatic dysfunction and intrahepatic cholestasis (Herrick, 1910; Green et al, 1953; Ahn et al, 2005). Choledocholithiasis and iron overload contribute to increased predisposition to extrahepatic cholestasis. Increased susceptibility to blood-borne hepatotropic viruses from numerous blood transfusions starting at an early age also contribute to the damage (Badawy et al., 2016). The clinical presentation can be acute liver involvement or chronic hepatobiliary manifestations (Shah et al, 2017). Acute liver involvement may present as acute sickle cell hepatic crisis, acute hepatic sequestration or acute intrahepatic cholestasis with rapid progression to acute liver failure. Chronic sickle cell hepatopathy can be secondary to acute hepatopathy or caused by an underlying constant injury to the liver.

Together with clinical presentation, a number of laboratory investigations and imaging can lead a physician to the diagnose SCLD. However, even decades from its first description in the scientific literature, liver biopsy remains the sole gold standard diagnostic tool for SCLD. Common histopathological findings in the liver of SCD patients are sinusoidal dilatation, Kupffer cell hyperplasia, erythrophagocytosis, focal necrosis, portal fibrosis and micronodular cirrhosis (Song, 1957). The severity of fibrosis and cirrhosis is thought to worsen with age as a consequence of accumulative insult.

Liver biopsy doesn’t come without risks: it is an invasive, painful and expensive diagnostic tool and can cause pain, haematomas, intraperitoneal haemorrhage, haemothorax, pneumothorax, bile leak causing peritonitis and transient bacteraemia, etc. (Boyum et al., 2016). These risks make it an impractical and near futile tool for progression or follow-up of liver disease in patients with SCD. It is known that patients with SCD are more susceptible to risks of liver biopsy than the general population regarding complications, especially during acute crises (Zakaria et al., 2003). Clinicians are left with the use of various non-invasive investigations to estimate the presence and extent of liver damage in SCD. These tools have been validated for other liver diseases but haven’t been evaluated for their performance in SCLD. Thus, there is an ongoing search for a newer, less invasive diagnostic tool as a better indicator of diagnosing underlying liver injury in patients with SCD.

We present a unique cohort of 50 patients with suspected SCLD. These patients were suspected to have underlying liver disease from their clinical presentation and elevated liver-related enzymes. They were referred for liver consultation by the primary Haematology team. Unexpectedly, only 10% (5 out of 50) of these patients had significant liver fibrosis (IF 3–6) on liver biopsy and only 3 out of those 5 had liver cirrhosis (IF 6). This indicated the limited value of traditionally used liver-related markers as a screening tool for the presence of SCLD. In search of a new diagnostic tool that was less invasive than a liver biopsy (percutaneous/ transjugular), we performed VCTE on all our patients. We were able to compare the results of VCTE to various findings within the liver tissue. Reliability of VCTE results may be influenced by operator experience, obesity, age, gender and presence of active inflammation (Castera et al., 2010). Although beyond the aim of the current study, if validated in SCLD, VCTE can become a useful tool in diagnosing the severity of liver fibrosis and reduce the need for invasive liver biopsies.

When grouped according to the severity of liver fibrosis (IF score on liver biopsy) the group with severe liver disease showed higher VCTE readings. P value and Spearman correlation rank showed strong significant positive correlation between TE and the two biopsy tissue determinants of liver fibrosis: IF score and HAI fibrosis score, weak but significant correlation was shown between TE and HAI inflammation score. Tissue iron scores, liver-related enzymes, such as ALT and AST, as well as platelet count did not correlate strongly or significantly with TE. These laboratory tests have been evaluated as indirect markers of hepatic fibrosis and cirrhosis in various liver diseases (Williams & Hoofnagle, 1988; Pilette et al., 1999; Wai et al., 2003; Sterling et al., 2006). However, in SCD, these tests and indices are affected by extra-hepatic processes, such as haemolysis and autosplenectomy, rendering them less specific; hence, not considered valuable determinants of hepatic injury (Rees et al, 2010). A significant number of patients with SCD have iron overload in their liver due to a history of multiple blood transfusions (Adamkiewicz et al., 2009). Markers of iron overload have been shown to significantly correlate with TE in the past (Drasar et al., 2017). We suspected that iron accumulation in the liver might cause an increase in TE, but this wasn’t the case in our cohort. All three of the liver tissue iron determinants (iron score, quantification and index) used showed no correlation with VCTE readings, even though the liver tissue iron determinants were significantly higher in the group with advanced fibrosis. Unfortunately, there was lack of availability of chelation data on our patients, limiting our conclusions about absence of correlation between TE and liver iron determinants.

We used measures of statistical analysis such as sensitivity, specificity and predictive values (PPV and NPV) to determine the performance of VCTE as a test of liver fibrosis with respect to the gold standard liver biopsy. We used two different points for the upper limit of normal (ULN) cut-off of elastography from VCTE. We chose 7·5 kPa as the first cut-off as it is an established point of reference for viral hepatitis. The second cut-off used, 8·5 kPa, was driven by our data that showed the highest sensitivity, specificity, PPV and NPV. Taking 8·5 kPa as the ULN for elastography resulted in sensitivity and NPV of 1, with higher specificity and PPV. When 7·5 kPa was used, as the cut-off, the number of false positives increased. We postulate that patients with SCLD have higher baseline elastography results than the viral hepatitis population. This can be attributed to chronic changes in liver parenchyma, congestion and recurrent episodes of inflammation. Our results suggest that TE can be used as a valuable tool for the evaluation of liver fibrosis in patients with SCLD despite significant hepatic iron accumulation. It is particularly a good screening modality to rule out SCLD rather than diagnose or accurately quantify the extent of tissue fibrosis. Therefore, defining a specific exclusion cut-off for TE on VCTE in this unique population of patients with SCD might reduce the need for liver biopsies in the future evaluation of liver disease.

The strength of our study lies in the availability of the gold standard liver biopsies for all our patients to diagnose and score for SCLD. Evaluation of all the biopsies by a single senior pathologist blinded to results of VCTE removed measurement as well as observer bias. Our study also has the highest number of patients enrolled with respect to past studies that evaluated the efficacy of TE in SCLD (Koh et al., 2013; Drasar et al., 2017; Pinto et al., 2017; Delicou et al., 2018). Our study uncovers limited value of traditional liver-related serology for diagnosing SCLD, as reflected by the low number of patients (10%) who actually had significant fibrosis on liver biopsy from our cohort of patients suspected to have SCLD.

We acknowledge that our small sample size, with a relatively small number of patients with advanced fibrosis and thus impaired distribution within the IF spectrum, are the drawbacks of our study. However, it is the largest report of liver biopsies in patients with SCD to date, and therefore an important contribution to the current knowledge.

Our findings should also be further evaluated with a validation cohort of patients with suspected SCLD, which could not be completed on this current study due to the small number of patients with advanced fibrosis.

Our study emphasizes the unmet need for further research on non-invasive tools and their utility in liver fibrosis evaluation in patients with SCLD.

In conclusion, VCTE may be used as an exclusion tool for diagnosis and follow-up in patients with SCLD. It may also play a role in the staging of Sickle Cell Hepatopathy in the future but should be evaluated by further studies. Systematic research is required for validation of VCTE in SCD. This can decrease the number of patients undergoing liver biopsies and might affect the current clinical practice paradigm for the evaluation of liver involvement in patients with SCD.

Table III.

Comparison of Transient Elastography with various markers of liver disease.

Variables (versus TE)	Spearman’s rank-order correlation	Confidence Interval	P-value
Ishak score biopsy	0·68	0·40 to 0·8	<0·0001
HAI fibrosis score	0·64	0·44 to 0·78	<0·0001
HAI inflammation score (0–18)	0·44	0·17 to 0·64	0·001
Iron score	0·27	−0·017 to 0·52	0·06
Iron, liver	0·157	−0·14 to 0·43	0·28
Iron index	0·07	−0·22 to 0·35	0·6
Serum iron	0·31	0·02 to 0·55	0·03
Ferritin	0·3	0·018 to 0·54	0·03
ALT	0·026	−0·26 to 0·31	0·86
AST	0·29	0·006 to 0·53	0·04
Platelet count	−0·3	−0·54 to −0·01	0·03

ALT, alanine aminotransferase; AST, aspartate aminotransferase; HAI, histology activity index; TE, transient elastography.