Background
Hepatocellular carcinoma (HCC) is a primary cancer of the liver that most often develops in identifiable patients with underlying liver disease, like hepatitis. HCC is the fourth most common cancer in the world with age-normalized incidence rates of 2.1 per 100,000 in North America [
1,
2]. The risk of HCC is thought to be associated with inflammatory changes within the liver microenvironment which extends over a protracted period of time, with cirrhosis and chronic hepatitis (B and C) accounting for approximately 50 % of all HCC [
3]. At our institution, we have a patient population with increased risk that present for treatment before emergence of HCC. This puts our institution in a unique position to evaluate biomarkers of cancer risk along with novel diagnostic tests that would improve stratification in patients with the highest likelihood of cancer, thus resulting in improved diagnostic and treatment paradigms to be applied. Our ability to impact patient outcomes in HCC rests in three key areas: 1) biomarkers of cancer risk, 2) improvements in diagnostic imaging, and 3) improved therapeutic options. This manuscript focuses on the first two issues and investigates the use of lysophosphatidic acid (LPA) profile analysis and PET/CT imaging in the multiple drug resistance-2 (MDR-2) knockout mouse model of HCC.
MDR2 is a membrane-associated protein linked to lipid transportation and is increased in HCC cell lines and tumors [
4,
5]. Previous studies have demonstrated
in vitro MDR2 expression is increased in response to chemotherapeutic agents and that HCC develops spontaneously in MDR2
−/− mice [
6‐
11]. More recently we have shown that HCC tumor burden can be reduced by administration of the commercially available LPA biosynthesis/signaling inhibitor BrP-LPA [
6] and confirmed the use of MDR2
−/− mice as a model for clinical pathologies [
12‐
16]. This current study builds on our use of macroscopic measurements of HCC as indicators of disease progression, and evaluates non-invasive markers and imaging in MDR2
−/− mice in order to validate tools pursuant to accurately measuring response to chemotherapeutic agents in future studies.
Currently, the first line therapy for patients not eligible for resection or liver transplantation is sorafenib chemotherapy [
17]. However, sorafenib is associated with a meager improvement in overall survival compared to supportive care alone. Recent reports have shown resistance to sorafenib in liver cells linked to phenotypic changes consistent with advanced invasion [
18], and that sorafenib response is related to deregulation of mitochondria fusion-related protein optic atrophy 1 (OPA1) and reduction in Phosphatase and tensin homolog (PTEN) expression [
19,
20]. The documented stable disease rate associated with sorafenib suggests that it would perhaps be more appropriate as a chemoprevention agent rather than treating established disease. To facilitate these studies an accurate suite of modalities to evaluate pre-clinical therapeutic response
in vivo is required, including methods to evaluate tumor burden and response to treatment. Accordingly, to better understand the MDR2
−/− model of HCC, MDR2
−/− mice underwent testing for: 1) HCC biomarker serum alpha fetoprotein (AFP) ELISA; 2) oxidative metabolism by
11C-acetate PET/CT; 3) glycolytic metabolism by
18 F-FDG PET/CT; 4) lipid metabolism by lysophosphatidic acid variant profile tandem mass spectroscopy; 5) cellular signaling by circulating cAMP ELISA and 6) inflammatory cytokine modulation by hepatic TNFα ELISA. We found that all modalities, except glycolytic metabolism via
18 F-FDG PET/CT, differentiated mice with HCC from those without, thus demonstrating potential modalities to monitor HCC development and treatment in MDR2
−/− mice. In addition, to confirm the relevance of
11C-acetate PET/CT for HCC in mice
11C-acetate PET/CT was performed in 8 patients with suspected recurrent HCC following standard of care therapy. We found that 5 of 8 patients were true positives, 2 of 8 were true negatives, and one false negative patient. This information confirms the potential of
11C-acetate PET/CT in murine studies to develop novel and new therapies of treatment.
Discussion
Hepatocellular carcinoma is one of the leading causes of deaths associated with cancer, with greater than 700,000 new cases diagnosed annually, and more than 600,000 deaths worldwide attributed to HCC per annum [
41]. Among malignancies, HCC is the fourth most common cancer worldwide [
1,
2] with an age-normalized incidence rate of 17, 42, 46, 62, and 371 per 100,000 in the United States, Africa, European Union, South East Asia, and China, respectively [
42].
Provided this, there is a distinct need to improve diagnostic and treatment options available for patients with HCC [
43]. The potential benefits of
11C-acetate PET/CT are numerous. In patients with elevated AFP > 200 or rising AFP but without any measureable disease burden
11C-acetate opens up the possibility of detecting occult or subclinical cancer within a high risk diseased liver. In patients with a prototype lesion(s) based on CT or MRI imaging that have addition lesions of unclear clinical significance because they are sub-centimeter
11C-acetate imaging opens up the possibility to put sub-centimeter lesions in perspective in patients who otherwise meet imaging criteria for HCC, thereby improving our ability to evaluate overall disease burden. In patients who meet transplant eligibility for transplant with a diagnosis of cancer based on AASLD criteria, but in whom our current metastatic evaluation is lacking both sensitivity and specificity
11C-acetate imaging will facilitate improved organ allocation and reduce post-transplant recurrence. In patients in whom clinical variables are worrisome for a greater disease burden than that suggested by our current imaging standards (CT or MRI) (i.e. acute elevation in AFP, portal vein thrombosis, multifocal lesions, tumor size > 4.5 cm)
11C-acetate opens up the possibility of improving our understanding of disease burden. In patients transplanted with cancer in whom the explant shows a greater disease burden than that noted on pre-transplant imaging putting them at a higher risk for cancer recurrence
11C-acetate imaging may provide some advantage over CT or MRI when used for surveillance in post-transplant patients considered at higher risk for recurrence and in whom clinical trials of novel immunosuppressant protocols post-transplant would become relevant. The MDR2
-/- mouse model of HCC provides a murine model whereby new paradigms of HCC diagnosis and targeted therapy can be evaluated [
6].
The purpose of this manuscript is to evaluate modalities of PET/CT for monitoring liver disease and HCC in MDR2
−/− mice in order to facilitate future studies aimed at exploring experimental treatments. Previous experiments using MDR2
−/− mice have relied upon macroscopic measurements of tumor burden [
6]; however, using this approach definitive assessments are only feasible at study termination, and therefore do not provide continual monitoring of disease progression and therapeutic response during the emergence and growth of HCC.
In general, monitoring of HCC is challenging, and diagnosis occurs when tissue fibrosis and HCC are advanced. However, HCC emerges in a well-defined cohort of patients with,
inter alia, Hepatitis C infection (HepC), non-alcoholic steatohepatitis (NASH) or alcoholic liver disease (ALD) cirrhosis, providing an identifiable HCC high-risk group. At Indiana University Hospital patients with HepC, NASH or ALD are generally monitored twice annually for HCC by measurement of AFP, liver biopsy histology, and ultrasonography. Advances in imaging have moved away from the use of ultrasonography and biopsies in favor of PET/CT imaging [
37‐
40,
44]. Moreover, the use of AFP in detection of HCC has proven problematic because of the low sensitivity and selectivity [
45]. Given the aforementioned limitations, this manuscript investigates PET/CT analysis of HCC along with serum/tissue biomarkers.
The first objective of this study was to examine current methods of HCC detection, serum AFP and PET/CT, in MDR2
−/− mice in order to illustrate compatibility and reproducibility to that seen in patients with HCC. Previous studies have shown that AFP is reduced in response to HCC treatment and that the level of this reduction is predictive of survival [
46]. Moreover, PET/CT is used to diagnosis HCC and follow response to treatment [
37‐
40]. To date HCC development has not been demonstrated in MDR2
−/− using either AFP or PET/CT.
We found that AFP generally increased contemporaneously with HCC levels in MDR2
−/− mice consistent with clinical studies [
47], and showed moderate sensitivity and selectivity for HCC. Consequently, serum AFP levels are a reasonable biomarker for HCC in MDR2
−/− mice and potentially could be used as a treatment response marker. However, there was a 30 % false negative rate, which argues that additional corroborate evidence would be required. In the clinic, AFP is very helpful for diagnosis and follow-up but negative AFP does not rule out the presence of HCC.
In the current work,
11C-acetate PET/CT imaging was shown to be a sensitive marker for monitoring HCC in MDR2
−/− mice. This was also seen with patients monitored for HCC reoccurrence [
39,
40]. In MDR2
−/− mice there was a significant increase in hepatic
11C-acetate metabolic rate when compared to controls. By comparison, there was no corresponding increase in hepatic
18 F-FDG up-take in mice. This observation is similar to previous work comparing
18 F-FDG, 6-Deoxy-6-
18 F-Fluoro-D-Glucose (
18 F-6FDG) and
11C-acetate PET/CT imaging in a hepatitis viral infection-induced woodchuck model. Using tumor to liver uptake ratios, tumors were detected in 53 % of animals using
18 F-FDG, 0 % using
18 F-6FDG and 94 % using
11C-acetate [
48]. These data are consistent with clinical observations of Cheung et al. which demonstrated that
11C-acetate was far superior to
18 F-FDG for HCC detection and staging in patients prior to liver resection or transplant [
39].
Interestingly, the false negative rate of
18 F-FDG in MDR2
−/− mice is consistent with clinical evaluation of
18 F-FDG as the primary imaging biomarker in HCC in patients [
37‐
40]. Moreover, recent quantitative studies of metabolic genes obtained from liver biopsies suggests that
18 F-FDG uptake in HCC is linked to expression of acetyl Coenzyme-A (CoA)-Synthetase-1 (ACSS1) an enzyme that generates acetyl-CoA from acetate and CoA, with the hydrolysis of ATP to AMP and diphosphate [
49,
50]. In mice hepatic ACSS1 expression is relatively low with respect to other acetyl-CoA enzymes and in the absence of other acetyl-CoA synthesizing enzymes ACSS1 is not sufficient to support murine development [
51]. The increase in cardiac
18 F-FDG in MDR2
−/− mice is not currently well understood, but may be linked to an increase in glucose transporters of the heart. It is well know that GLUT4, the main contributor of glucose transport in the heart [
52] and MDR2 are both regulated by PI3K/AKT signaling pathway. PI3K-AKT controls the GLUT4 translocation from intracellular vesicles to the plasma membrane [
53]. In a similar manner PI3K/AKT activity also regulates MDR2 transporter activity [
54]. At this time no information exists pertaining to cardiac AKT levels in MDR2
−/− mice or the effect of MDR2 gene deletion on cardiac GLUT4 expression; however, current reports show that reduction of MDR activity with functional inhibitors increases chemo-sensitivity without affecting AKT phosphorylation [
55]. The connection amongst MDR2, GLUT4, and
18 F-FDG in the heart requires additional investigation to better understand any interactions.
We are cognizant that observations in MDR2 gene knockout mice may not be 100 % comparable to that seen in patients. In MDR2
−/− mice chemotherapeutic clearance is potentially impaired, which would provide for a sustained dosing time [
56]. MDR2
−/− mice have a significant impairment of
99mTc-Sestamibi biliary excretion [
57], which is a marker of MDR function, and thus biliary transport is reduced in MDR2
−/− mice. However, this does not appear to interfere with
11C-acetate PET/CT imaging which was clearly elevated in MDR2
−/− when compared to controls. To mitigate possible hepatobiliary change in
11C-acetate transport, standardized protocols could be used where treated and untreated groups are imaged longitudinally, thus increasing the statistical power and minimizing the possibility of erroneous results.
The second objective of this manuscript was to better understand the connection between lysophosphatidic acid (LPA) and HCC in MDR2
−/− model. Previous studies have identified the biosynthesis of LPA as a marker of HCC [
58‐
60], and is known to be associated with tissue fibrosis [
61‐
63] in this patient population. Moreover, we have previously shown that LPA variants were able to differentiate between liver disease and patients with HCC [
59]. Interestingly, these data are consistent with our current, and previous [
6], studies in MDR2
−/− mice where LPA is not only elevated contemporaneous with HCC, but the LPA 20:4 and 18:2 variants are highly consistent with data from the clinical populations with HCC. Currently the precise role and mechanism connecting LPA to HCC is not fully understood. However, LPA is known to be associated with endothelin derived growth receptors (EDGR) [
64], and has previously been linked to cellular signaling molecules (cAMP) and pro-inflammatory cytokines (TNFα) [
65,
66]. In MDR2
−/− mice with HCC there was a significant decrease in serum cAMP contemporaneous with an increase in LPA. The LPA receptor-3 (LPA3) is reported to be inhibitory of adenylate cyclase activity [
64], and mice deficient in adenylate cyclase are desensitized to LPA [
67]. Interestingly, reductions in cAMP have previously been reported in HCC [
68]. Moreover, LPA has also been linked to the elevated expression of TNFα [
69] which has been shown in patients with HCC [
70]. In MDR2
−/− mice hepatic TNFα levels were increased concurrent with HCC and increased LPA. Combined, these data suggests that elevated LPA biosynthesis associated with HCC is potentially linked to modulations of cAMP and/or TNFα. These data clearly illustrate the multitude of pathways by which LPA can exert action within the oncogenic milieu. These interactions are exemplified by,
inter alia, advanced glycation end products (RAGE) and hypoxia inducible factor 1 alpha (HIF1α) which are known to modulate cellular signaling associated with a variety of cancer types [
71,
72]. In addition to binding to EDGR receptors, LPA has also been shown to bind to the receptor for RAGE [
73], which is a key regulator of inflammation-associated liver carcinogenesis in MDR2
−/− mice [
71]. Similarly, HIF1α has been shown to suppress apoptosis in HCC cells [
72] and increases cellular responsiveness to LPA [
74]. Taken overall, these data suggest that LPA plays a key role in the development and progression of HCC in MDR2
−/− and exerts its effects via a range of second messenger systems and/or modulation of cytokines and growth factors.
LPA variant analysis complements
11C-acetate PET/CT imaging. In MDR2
−/− mice changes in the LPA profile are contemporaneous with the emergence of HCC and in patients the LPA variant profile is corrected by liver transplantation [
59]. The analysis is sensitive (nmol.ml), standard, inexpensive and can be incorporated in to a routine monitoring program to indicate the need for
11C-acetate PET/CT. Following LPA variant analysis
11C-acetate PET provides fmol/ml sensitivity and spatial information in a patient population which has non-diffuse disease, this would allow for
90Y thearaspheres or stereotactic body radiotherapy to be deployed as treatment of HCC.
Authors’ contributions
NS and MM conceived of study, and participated in its design and coordination. NS and PT drafted the manuscript. PT supervised PET/CT imaging in mice and interpreted the results. AR and BM carried out the PET/CT imaging on mice. RS carried out and interpreted the histology. MT and JF carried out the PET/CT imaging on humans. NS performed/supervised LPA, TNFα, cAMP and AFP analysis. All authors read and approved the final manuscript.