Limitations of Fluorine 18 Fluoromisonidazole in Assessing Treatment-induced Tissue Hypoxia after Transcatheter Arterial Embolization of Hepatocellular Carcinoma: A Prospective Pilot Study

Abstract

Purpose

To determine the variance and correlation with tumor viability of fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) uptake in hepatocellular carcinoma (HCC) prior to and after embolization treatment.

Materials and Methods

In this single-arm, single-center, prospective pilot study between September 2016 and March 2017, participants with at least one tumor measuring 1.5 cm or larger with imaging or histologic findings diagnostic for HCC were enrolled (five men; mean age, 68 years; age range, 61–76 years). Participants underwent ¹⁸F-FMISO PET/CT before and after bland embolization of HCC. A tumor-to-liver ratio (TLR) was calculated by using standardized uptake values of tumor and liver. The difference in mean TLR before and after treatment was compared by using a Wilcoxon rank sum test, and correlation between TLR and tumor viability was assessed by using the Spearman rank correlation coefficient.

Results

Four participants with five tumors were included in the final analysis. The median tumor diameter was 3.2 cm (IQR, 3.0–3.9 cm). The median TLR before treatment was 0.97 (IQR, 0.88–0.98), with a variance of 0.02, and the median TLR after treatment was 0.85 (IQR, 0.79–1), with a variance of 0.01; both findings indicate a narrow range of ¹⁸F-FMISO uptake in HCC. The Spearman rank correlation coefficient was 0.87, indicating a high correlation between change in TLR and nonviable tumor.

Conclusion

Although there was a correlation between change in TLR and response to treatment, the low signal-to-noise ratio of ¹⁸F-FMISO in the liver limited its use in HCC.

Keywords: Molecular Imaging-Clinical Translation, Embolization, Abdomen/Gastrointestinal, Liver

Clinical trial registration no. NCT02695628

© RSNA, 2022

Summary

Although hypoxia imaging after embolization treatments for hepatocellular carcinoma may help predict short-term recurrence, the low signal-to-noise ratio of fluorine 18 fluoromisonidazole likely limits its use for tumor hypoxia imaging in the liver.

Key Points

• ■ Fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) PET/CT indicated low variance in ¹⁸F-FMISO uptake in hepatocellular carcinoma (HCC) before and after transcatheter arterial embolization.

• ■ There was statistical correlation (Spearman rank correlation coefficient, 0.87) between changes in tumor-to-liver (TLR) ratio for ¹⁸F-FMISO uptake and HCC recurrence.

• ■ ¹⁸F-FMISO as an imaging agent for hypoxia had an undesirably low signal-to-noise ratio in the liver, with a median TLR of 0.97 before treatment and 0.85 after treatment.

Introduction

Primary liver cancer is the fastest-growing cancer (in rate) in the United States, and more than 42 000 cases were estimated to be diagnosed in 2021, almost three-fourths of which were hepatocellular carcinoma (HCC) (1). The estimated 5-year survival rate is 20%. Worldwide, primary liver cancer represents a greater burden; HCC is the second most common cause of cancer-related deaths (2). Local-regional intra-arterial therapy is a mainstay of treatment (2) and includes transcatheter arterial chemoembolization (TACE) with an ethiodized oil emulsion, drug-eluting bead TACE, small particle transcatheter arterial embolization (TAE), and yttrium 90 (⁹⁰Y) radioembolization. Each of these therapies involves deposition of small particles directly into tumor vasculature via selective catheterization of hepatic, and occasionally extrahepatic, arteries, thereby inducing changes in the tumor microenvironment.

The mechanism of action of TAE, drug-eluting bead TACE, and TACE within the tumor microenvironment is in part through induction of hypoxia, with smaller particle size creating increased hypoxia (3). Although hypoxia has been shown to lead to apoptosis, it may also reduce therapeutic effectiveness of radiation and chemotherapy because increased expression of hypoxia-inducible factor 1-α correlates with poorer survival across a number of different cancers (4). Hypoxia-inducible factor 1–driven upregulation of growth factors has been associated with poor prognosis in untreated HCC (5–7). Unfortunately, few data are available on how these changes affect response, recurrence, metastatic potential, and treatment resistance. As a result, more effective real-time monitoring has been suggested to optimize treatment and follow-up strategies (8). Therefore, quantifying hypoxia before and after treatment may be a prognostic marker for HCC recurrence that can help spur the development of follow-up strategies. Given that intra-arterial treatment at CT or MRI typically is not evaluated until 2–3 months after the procedure (9), there remains an unmet clinical need for earlier evaluation in which hypoxia imaging could be performed within 24 hours after the procedure to determine the effect of treatment.

The radioisotope fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) has been shown to accumulate in hypoxic areas of the liver after surgical ligation of right and left hepatic artery branches in a porcine model (10). The radiotracer and its metabolites enter cells by passive diffusion, where it undergoes reduction. The radiotracer cannot undergo reoxidation in hypoxic conditions; therefore, it remains trapped within the cell (11). ¹⁸F-FMISO has been shown to have predictive value in response to radiation therapy in several tumors, including lung cancer, head and neck cancer, renal cell cancer, and gastrointestinal cancers (12–16). There is particular interest in using hypoxia imaging in these tumors because hypoxia limits the effectiveness of radiation therapy, particularly in lung cancer, and so-called radiation dose painting on the basis of hypoxia imaging could be used to modulate dose based on the presence of hypoxia within a tumor (17). Although ¹⁸F-FMISO has been studied in other solid tumors, to our knowledge, it has not been previously studied in HCC. Thus, it is unknown whether ¹⁸F-FMISO could be used to determine where areas of hypoxia exist after TAE or TACE to help predict response to treatment. Many other hypoxia agents exist, including ¹⁸F-EF5, ¹⁸F-fluoroazomycin arabinoside, ¹⁸F-HX4, and ⁶⁴copper–labeled diacetyl-bis(N4-methylthiosemicarbazone), but, to our knowledge, none have been as widely studied as ¹⁸F-FMISO (18). In addition, the cost of manufacturing other agents is substantially higher, which makes ¹⁸F-FMISO a reasonable choice for use in a pilot investigation for HCC.

The purpose of our prospective pilot study was to determine the variance of ¹⁸F-FMISO uptake in HCC tumors compared with normal liver before and after TAE and to determine if any correlation with recurrence exists.

Materials and Methods

This investigator-initiated pilot clinical trial was registered on the ClinicalTrials.gov website before participant enrollment (ClinicalTrials.gov identifier: NCT02695628) to reduce citation bias. Institutional review board approval was obtained. All procedures were performed in accordance with the Good Clinical Practice guideline and the Health Insurance Portability and Accountability Act.

Study Participants

This was a single-arm, single-center, prospective pilot study. Between September 2016 and March 2017, patients with HCC diagnosed at imaging or biopsy who were referred to the interventional radiology clinic were prospectively screened and enrolled in this study. Informed consent was obtained for all participants. Participants were deemed eligible if they were 18 years or older and had an Eastern Cooperative Oncology Group performance status score of 0 (fully active), 1 (restricted in strenuous activity), or 2 (ambulatory but unable to work, up more than 50% of waking hours); Child-Pugh-Turcotte category A (preserved hepatic function) or B (moderately impaired hepatic function); and imaging and clinical features of tumor diagnostic for HCC according to the Liver Reporting and Data System (19). A maximum of five participants were targeted for enrollment on the basis of availability of the radiopharmaceutical. The full list of inclusion and exclusion criteria, which was developed to reduce selection bias, is provided in Table 1.

Table 1:

Inclusion and Exclusion Criteria

Data Collection

All participants’ cases were presented to the institutional multidisciplinary liver tumor board, and those deemed suitable for intra-arterial therapy were included. Enrolled participants were required to undergo contrast-enhanced triphasic CT or multiphasic MRI within 8 weeks of initial treatment. Participants were also required to have undergone ¹⁸F-FMISO PET/CT imaging within 4 weeks before treatment to obtain baseline measurements of ¹⁸F-FMISO uptake. Participants then underwent TAE. Studies (3,20) suggested increasing hypoxic fraction in the tumor with smaller particle use; therefore, 40–120 µm trisacryl gelatin microspheres (Embosphere; Merit Medical) were selected for primary use, with 100–300 µm spheres used for secondary embolization as needed. Neither glass nor resin radioembolization ⁹⁰Y therapies were used because pair production from minor positron decay could affect ¹⁸F-FMISO PET camera counts. Participants underwent repeat ¹⁸F-FMISO PET/CT between 16 hours and 24 hours after completion of treatment. All participants were followed up with repeat triphasic CT or MRI 2 months after treatment, with continued follow-up imaging if there was no evidence of disease, or treatment as required if there was residual or recurrent disease, for up to 6 months after initial treatment. Follow-up was performed at scheduled in-person clinic visits to reduce recall bias. Adverse events were reported using the Common Terminology Criteria for Adverse Events (version 5.0; https://ctep.cancer.gov/protocoldevelopment/electronic_applications/ctc.htm#ctc_50) of the National Cancer Institute.

Imaging Protocols

After participants were successfully screened and enrolled, triphasic CT or multiphasic MRI was performed unless the participant had already undergone triphasic CT or multiphasic MRI within 8 weeks of embolization treatment. Baseline ¹⁸F-FMISO imaging was performed with a PET/CT scanner (Discovery; GE Healthcare) before TAE. ¹⁸F-FMISO was prepared by experienced radiopharmaceutical chemists (F.T.C.) from the affiliated cyclotron and radiochemistry facility. On the day of the scan, 10 mCi of ¹⁸F-FMISO was injected intravenously into the participant. PET/CT imaging was performed between 90 and 150 minutes after injection. All PET/CT imaging was performed using the same GE Discovery scanner with the following parameters: 120 kV; automatic tube current (range, 30–100 mA); noise index, 25; and rotation time, 0.5 second. The vital signs in all participants were monitored for up to 3 hours after injection for evidence of any reaction. Images were acquired and uploaded to the picture archiving and communication system. No home medications were discontinued for the purposes of the scan.

TAE Procedure

TAE was performed at least 10 half-lives, or 18 hours, after injection of ¹⁸F-FMISO. Embolization was performed by using superselective technique with small particles until complete stasis (ie, no antegrade flow after five cardiac cycles) was achieved. Initial embolization was performed with one vial of 40–120 µm particles; if the vial was completed, additional embolization was performed with 100–300 µm particles. All embolization procedures were performed by a single interventional radiologist with 7 years of experience (R.P.S.) to reduce performance bias. At completion of the procedure, the participants were admitted for overnight observation with repeat ¹⁸F-FMISO imaging the next day, as described previously. For residual and/or recurrent tumors discovered during follow-up, additional treatments using any available option were offered.

Image Analysis

¹⁸F-FMISO image interpretation was performed by a single nuclear medicine physician with 31 years of experience (G.M.S.) by drawing a spherical volume of interest 3 cm ± 1 in diameter over a region of liver unaffected by tumor. On scans acquired after treatment, the volume of interest was drawn in areas that did not undergo treatment. The average standardized uptake value (SUV) of this operator-defined volume of interest was determined. This was repeated in two additional areas of unaffected liver, and the mean of these three values was used as the SUV_mean of the normal liver. A second volume of interest encompassing a sphere measuring 1 cm in diameter (to ensure consistency) was drawn at the center of the index tumor or tumors, each of which was greater than or equal to 1.5 cm in size. The maximum SUV, or SUV_max, was calculated from this region, with the threshold set at 10% of peak SUV of all voxels in the tumor volume. The tumor-to-liver ratio (TLR) was calculated as follows, based on prior studies: TLR = SUV_max tumor/SUV_mean liver (21). The TLR was used instead of SUV_max to ensure that comparisons between participants were normalized on the basis of liver activity.

Statistical Analysis

Before enrolling participants, it was determined that for 80% power and an α value of .05, assuming an SD of 1 for each mean TLR value for five participants, the minimum detectable difference between the means was 1.77. Because no previous data were available, we recognized that it was unlikely that the difference in means would be high enough to detect in this small pilot study.

Statistics were calculated on a per-tumor basis, because only one participant had two tumors, and the other participants had one tumor each. Median, IQR, mean, variance, and SD are provided for summary statistics. Statistical analysis comparing the difference in mean TLR before and after embolization was performed by using the Wilcoxon rank sum test. A P value of .05 or less was considered to indicate statistical significance. For determining correlation with recurrence, a Spearman rank correlation coefficient was calculated. All statistical analyses were performed by using RStudio software (version 1.4.1106; R Foundation for Statistical Computing).

Results

Participant Characteristics

Figure 1 shows the flowchart of study participants. Six participants were prospectively screened and five met enrollment criteria (mean age, 68 years; age range, 61–76 years; all men). One participant had an Eastern Cooperative Oncology Group performance status score of 0, and four had an Eastern Cooperative Oncology Group performance status score of 1. Four participants were classified as Child-Pugh-Turcotte category A, and one was classified as Child-Pugh-Turcotte category B. The mean α-fetoprotein level was 114 ng/mL (range, 10.3–312.4 ng/mL), and mean tumor volume was 15.1 cm³ (range, 5.1–24.8 cm³).

Figure 1:

Participant study flowchart. Participants were screened in the interventional radiology clinic after consent was obtained and underwent fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) PET/CT imaging if criteria were met.

One participant dropped out of the study prior to undergoing repeat ¹⁸F-FMISO imaging and was therefore not included in the analysis (Fig 1).

Comparison between TLR before and after Treatment

Table 2 shows the tumor data for the four participants included in our analysis. The median tumor diameter was 3.2 cm (IQR, 3.0–3.9 cm). The TLR before versus after treatment for each participant is plotted in Figure 2. The median TLR before treatment was 0.97 (IQR, 0.88–0.98), with a variance of 0.02; the median TLR after treatment was 0.85 (IQR, 0.79–1), with a variance of 0.01. We found no evidence of a difference in mean TLR before and after TAE treatment (0.98 ± 0.15 vs 0.85 ± 0.12; P = .44), in part because of a decrease in TLR in viable tumor and an increase in TLR in nonviable tumor. Tumors with viability after embolization all showed a decrease in pretreatment to posttreatment TLR of 0.07, 0.43, and 0.04 on the initial 2-month posttreatment triphasic CT or MRI; conversely, the TLR increased by 0.02 and 0.06 in two participants with nonviable tumor. Figure 3 shows posttreatment CT and ¹⁸F-FMISO PET/CT scan results for participant 1 (viable tumor), and Figure 4 shows posttreatment CT and ¹⁸F-FMISO PET/CT scan results for participant 3 (nonviable tumor). Increase in TLR was highly correlated with nonviable tumor, with a Spearman rank correlation coefficient of 0.87.

Table 2:

Participant Tumor Data Before Embolization

Figure 2:

Tumor-to-liver ratio (TLR) plotted for tumor viability before treatment versus after treatment. The TLR after embolization was higher for nonviable tumors than for viable tumors.

Figure 3:

Fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) uptake in the liver before and after embolization for participant 1 (66-year-old man with hepatocellular carcinoma) with residual tumor after treatment. (A) Axial contrast-enhanced CT image acquired before embolization shows enhancing tumor (arrow) in segments 5/8. (B) Axial ¹⁸F-FMISO PET/CT image acquired before embolization shows uptake in the tumor (arrow) similar to that of liver parenchyma, with yellow and red areas indicating increased uptake. (C) Axial ¹⁸F-FMISO PET/CT image acquired after embolization shows minimal uptake in the tumor (arrow). (D) Axial contrast-enhanced CT image acquired 2 months after embolization shows enhancing viable tumor (arrow).

Figure 4:

Fluorine 18 (¹⁸F) fluoromisonidazole (FMISO) uptake in the liver before and after embolization in participant 3, a 65-year-old man with hepatocellular carcinoma and nonviable tumor after treatment. (A) Axial contrast-enhanced CT image acquired before embolization shows a segment-7 enhancing tumor (arrow). (B) Axial ¹⁸F-FMISO PET/CT image acquired before embolization shows uptake in the tumor (arrow) similar to that of liver parenchyma, with yellow and red areas indicating increased uptake. (C) Axial ¹⁸F-FMISO PET/CT image acquired after embolization shows continued uptake in the tumor (arrow). Note the focal contrast retention in the tumor from retained embolic particles. (D) Axial contrast-enhanced CT image acquired at 2 months after embolization shows necrosis with no viable tumor (arrow).

Adverse Events

No adverse events occurred during or immediately after administration of ¹⁸F-FMISO or performance of PET/CT examination. All adverse events occurred after the embolization procedure: Four participants experienced grade 1 adverse events (hepatic pain, nausea, diarrhea, fatigue, elevated liver enzymes), one participant experienced grade 2 adverse events (hepatic pain and cellulitis), and one participant experienced a grade 3 adverse event (hepatic pain). None of the participants experienced a Common Terminology Criteria for Adverse Events grade 4 or 5 event.

Discussion

The purpose of our study was to determine the variance of ¹⁸F-FMISO uptake in HCC versus normal liver before and after embolization and to determine if any correlation with recurrence exists. We found that variance was 0.01 before TAE and 0.02 after TAE, which demonstrates consistent ¹⁸F-FMISO uptake. We also found that the Spearman rank correlation coefficient was 0.87, which demonstrates a high correlation between ¹⁸F-FMISO uptake and recurrence. This study provides support that there may be a role for hypoxia imaging before and after HCC treatment.

Although comparison between pre- and posttreatment TLR demonstrated very little variance in the TLR, it also showed very little difference in ¹⁸F-FMISO uptake between the liver and tumor prior to any treatment, as demonstrated by the mean TLR of close to 1. As a result, any hypoxia in tumors may be masked by increased ¹⁸F-FMISO uptake by hepatic parenchyma. Indeed, there was a low signal-to-noise ratio on visual inspection of the images, and areas of hypoxia within tumors were not readily distinguishable on PET/CT images prior to any treatment.

Evaluation of the SUV_max of the tumor demonstrated a decrease in mean TLR after treatment, although the study was not powered to detect the small change. Tumors with an increase in TLR after treatment showed a complete tumor response at the initial 2-month posttreatment scan, whereas those with a decrease in TLR showed viable tumor. Thus, the Spearman correlation coefficient showed a high correlation between the change in TLR and nonviable tumor. These shifts in TLR were small, however, and it is important to note that it remains uncertain whether this is a true correlation, considering the small sample size.

Interestingly, our results demonstrate that tumors that spanned multiple segments remained viable, whereas tumors in single segments were nonviable. The reduced response rate of tumors in watershed regions of the liver (22,23) likely explains tumor viability and points to a possible reason for increased TLR in nonviable tumors on 2-month follow-up CT images. Lesions in a single segment of the liver may have been more likely to develop hypoxia because of the complete occlusion of the segmental branch, thus allowing for increased ¹⁸F-FMISO accumulation in the tumor and a higher TLR. The tumors in watershed areas, however, likely had continued perfusion from untreated segmental branches, and thus, may have been less likely to have hypoxic areas. Although it would be expected that the TLR change would not be significant in cases of viable tumor, both the quantitative values and qualitative appearance on PET/CT images seem to indicate lower ¹⁸F-FMISO uptake in lesions that do not respond. Determining the cause of this is difficult because of the high liver uptake of ¹⁸F-FMISO, which presents another limitation for use of this radiopharmaceutical.

A possible benefit of the use of hypoxia imaging agent after intra-arterial liver HCC therapy could be more rapid evaluation of the treatment effect. Currently, imaging is typically performed 1 or 2 months after intra-arterial therapy to evaluate for recurrence (9). Although some patients may undergo same-day discharge, in the United States the majority of patients are admitted for several days after treatment (24,25). During this time, performance of PET/CT examination may be feasible to evaluate for hypoxia radioisotope uptake. Several questions remain, however, including the optimal timing of imaging after treatment and the optimal hypoxia imaging agent. Because of the high hepatic parenchymal uptake, ¹⁸F-FMISO is unlikely to be an ideal imaging agent for this purpose. Other agents exist and may be better options for hepatic hypoxia imaging, including ¹⁸F-HX4 (26,27), which has shown lower uptake in the liver parenchyma. Before treatment, hypoxia imaging could be used before delivery of hypoxia-sensitizing agents such as tirapazamine. Recently, a phase I trial (28) found that tirapazamine combined with TAE for HCC had a safe adverse-effect profile in 27 participants and 60% of tumors had a complete response. Using a hypoxia tracer before or after tirapazamine use could help select patients who are most likely to benefit from treatment with a hypoxic sensitizer.

Our study had several important limitations. Only four participants underwent complete imaging and were analyzed; thus, it is difficult to make any firm conclusions based on the data. Nonetheless, it appears clear that ¹⁸F-FMISO presented a barrier for hepatic hypoxia imaging because of its background activity in the liver. A further limitation was the lack of a standard value to use as an objective measure of hypoxia. We used TLR in this study on the basis of tumor-to-blood and tumor-to-muscle ratios in head and neck cancer, lung cancer, and other solid tumors. ¹⁸F-FMISO is partially cleared by the liver and allows for better comparison of the signal-to-noise ratio between different patients with HCC. However, further determination in larger studies is needed to assess whether TLR is the best objective measure. An additional limitation was the uncertainty concerning the optimal imaging time using hypoxia tracers after embolization, which warrants further investigation if other radiopharmaceutical agents have lower background hepatic uptake. A recent study (3) suggested that maximum hypoxia occurs 3 days after embolization and that it increases with smaller particle size, independent of TAE or drug-eluting bead TACE. Thus, future studies may find improved hypoxia tracer signal-to-noise if imaging is performed at a later time; however, this may be inconvenient for the patient who may need to return for imaging after discharge. Finally, our study used small particle bland embolization without chemotherapy for several reasons: the availability of data on the effect of small bland particles on hypoxia in HCC (20,29), the effect of chemotherapy agents and ethiodized oil on ¹⁸F-FMISO uptake were unknown, and prior studies showed no difference in outcomes between TAE and drug-eluting bead TACE (30). It is unclear how the addition of chemotherapy or use of ethiodized oil may affect hypoxia imaging.

In conclusion, although this study provides initial data for the use of ¹⁸F-FMISO to study liver tumors, the radiopharmaceutical is limited in its use because of its high hepatic background uptake, as well as the lack of significantly greater uptake by HCC despite its low variance. Although hypoxia imaging of HCC may be feasible and eventually may play a role in early treatment evaluation, other radiopharmaceutical agents should be investigated.