Perfusion Change of Hepatocellular Carcinoma during Atezolizumab plus Bevacizumab Treatment: A Pilot Study

DOI: https://doi.org/10.21203/rs.3.rs-1722962/v1

Abstract

Purpose

To investigate whether the early perfusion change in hepatocellular carcinoma (HCC) predicts the long-term therapeutic response to atezolizumab plus bevacizumab.

Methods

We retrospectively selected 19 subjects (median age: 62 years, 4 females and 15 males) having advanced HCC and treated with atezolizumab alone (n = 3) or in combination with bevacizumab (n = 16). The 4-phased CT or MRI imaging was performed for each subject before and at 9 ± 2 and 21 ± 5 weeks after therapy initiation. The tumor-to-liver signal ratio in the arterial phase was used to estimate the tumor perfusion. The change in tumor perfusion from the baseline to the 1st follow-up exam was correlated with the tumor response evaluated using mRECIST at the 2nd follow-up exam. The difference between favorably responding and non-responding groups was statistically analyzed using one-way ANOVA.

Results

The mean tumor long axis in the baseline image was 59 ± 47 mm. The HCC perfusion changes were − 26 ± 18% for complete (or partial) response (CR/PR, n = 8), -24 ± 12% for stable disease (SD, n = 8), and 9 ± 13% for progressive disease (PD, n = 3). The HCC perfusion change of the CR/PR groups was significantly lower than that of the PD group (p = 0.0040). The HCC perfusion changes between the SD and PD groups were also significantly different (p = 0.0135). The sensitivity and specificity of the early perfusion change to predict the long-term progressive disease were 100% and 94%, respectively.

Conclusion

The early change in HCC perfusion may predict the long-term therapeutic response to atezolizumab plus bevacizumab, promoting personalized treatment for HCC patients.

Introduction

Hepatocellular carcinoma (HCC) is the fifth most common cancer worldwide and the third leading cause of global cancer-related death (1). Although liver transplantation and, to a lesser extent, resection offer potential for cure, most patients present with a disease that is too advanced for curative surgical therapy (2). Table 1 lists ten systemic therapeutic regimens approved by the federal drug administration (FDA) for the advanced HCC. The therapeutic agents can be largely categorized into three types: immune checkpoint inhibitors (PD-1 or CTLA-4 inhibitors), tyrosine kinase inhibitors, and angiogenesis inhibitors (VGEF or VGEFR2 inhibitors). Immunotherapy is to harness the patient’s own immune system to fight against cancer (3). Cancer cells exploit immune checkpoints to evade the immune-cell attack (4), thus immune checkpoint inhibitors resuscitate the host immune system against cancer cells. Atezolizumab, an immune checkpoint inhibitor, combined with bevacizumab has been recently adopted as the first-line treatment for the advanced HCC (5). The tyrosine kinase inhibitors, suppressing cell proliferation and angiogenesis, or the anti-angiogenic agent (ramucirumab), are used as the secondary options for advanced HCC patients when immune checkpoint inhibitors are ineffective (6).

Immune checkpoint inhibitors, however, are costly (7–9) and yield the clinical benefit only for a portion of cancer patients (10, 11). The objective response rate of atezolizumab plus bevacizumab for advanced HCC is 36% (5). Therefore, a reliable tool to assess its therapeutic response early is urgently needed. Currently, the HCC response to systemic therapy is assessed by the size change of viable tumors based on modified Response Evaluation Criteria in Solid Tumors (mRECIST) (12). However, the tumor size changes relatively slowly, so an accurate assessment of treatment efficacy often requires prolonged follow-ups.

Change in tumor perfusion may serve as a reliable surrogate biomarker to predict the HCC response to immune checkpoint inhibitors. Zheng et al. have recently demonstrated the significant increase of tumor perfusion via the CD8+ T-cell activation after effective immune checkpoint inhibition prior to the change in tumor size in orthotopic and spontaneous breast and colon cancer mouse models (13). Also, Shigeta et al. have reported that atezolizumab plus bevacizumab increased the microvessel density significantly more than atezolizumab monotherapy in two orthotopic HCC mouse models (14), when the dose of bevacizumab was adjusted for vascular normalization like the human dose (15). Therefore, these preclinical data suggest that the HCC perfusion, favorably responding to the atezolizumab plus bevacizumab, will likely be increased. However, the HCC is hypervascular and hyperperfused in general, and it is uncertain whether the perfusion will be further increased in human subjects after an effective atezolizumab plus bevacizumab treatment.

This pilot study primarily aimed to test whether the HCC perfusion will be increased after an effective atezolizumab plus bevacizumab, and second, whether the early perfusion change will predict the long-term therapeutic efficacy.

Materials And Methods

Patients

Figure 1. Flow chart of the subject-selection process. Among the 22 patients who entered systemic therapy, including atezolizumab (January 2020 – May 2021), three patients were deceased before the 2nd follow-up scan. Sixteen patients were treated with atezolizumab plus bevacizumab, while three patients were treated with atezolizumab alone due to the adverse side effect of bevacizumab.

This study was approved by the institutional review board at the University of Alabama Birmingham. Figure 1 shows the flow chart summarizing the subject-selection process. We retrospectively selected twenty-two therapy naïve patients having advanced HCC (seventeen males, five females) with a median age of 62 and age range of 37-87, who entered systemic therapy, including atezolizumab, at UAB (search period: January 2020 to May 2021). We excluded three subjects deceased before the second follow -up exam. All subjects were considered to be treated with atezolizumab plus bevacizumab, but three patients could not take bevacizumab because of adverse side effects. Among the nineteen subjects, twelve were Caucasians, five were African Americans, one was Asian, and one was Hispanic.

Imaging protocol

Standard 4-phase liver CT or MRI imaging (baseline, arterial phase, portal vein phase, and late phase) was carried out before treatment and every 2 ~ 3 months after starting therapy. Five CT scanners across UAB campus were used, including iCT 256 (Philips Healthcare, Amsterdam, Netherlands), Brilliance 64 (Philips Healthcare, Amsterdam, Netherlands), Revolution CT (GE Healthcare, Chicago, IL), Discovery CT750 HD (GE Healthcare, Chicago, IL), and SOMATOM Definition AS (Siemens Medical Solutions USA, Inc., Malvern, PA). All CT scans used 70 KeV X-rays. The pixel size, matrix size, and slice thickness were 0.6 ~ 0.9 mm, 512×512, and 2.5 mm, respectively. Four MRI scanners were used, including 3T Skyra (Siemens Medical Solutions USA, Inc., Malvern, PA), 1.5T or 3T Achieva (Philips Healthcare, Amsterdam, Netherlands), and 3T Ingenia (Philips Healthcare, Amsterdam, Netherlands). MRI images were acquired using fat-suppressed 3D spoiled gradient echo sequences with the following parameters: TR = 3.0 ~ 4.0 ms, TE = 1.3 ~ 1.9 ms, pixel size = 1.0 ~ 1.2 mm, matrix size = 320×320 ~ 350×350, slice thickness = 3 ~ 5 mm, flip angle = 9 ~ 10°, and the number of average = 1. Omnipaque (GE Healthcare, Chicago, IL) was used for CT imaging, while gadoteridol (Bayer healthcare Pharmaceutical, Wayne, NJ) was used for MRI. CT scans were used for nine subjects, MRI scans were used for seven subjects, and both CT and MRI scans were used for the other three subjects. The first and second follow-up scans were implemented at 9 ± 2 and 21 ± 5 weeks after treatment onset. The tumor response to therapy was assessed using the mRECIST criteria at the second follow-up exam.

Image Processing

A single image slice crossing the center of a tumor was selected, and the tumor boundary was delineated by a board-certified radiologist at the artery phase. In the presence of multiple tumors, up to the three largest tumors in the liver were selected. The liver tissue region near each tumor (~ 5 cm in diameter) was also selected. The CT (or MRI) pixel values within the tumor (or liver) region were averaged, and the tumor-to-liver signal ratio, defined as the mean tumor pixel value divided by the mean liver pixel value, was calculated. All CT pixel values were in the Hounsfield unit, whereas the MRI pixel values were in an arbitrary unit. Image processing was conducted using a free software package, ImageJ v1.53a (Bethesda, MD).

Statistical Analysis

All data in this manuscript are presented as means ± standard deviation (SD). Statistical difference between two groups was determined using one-way ANOVA (16). Pearson correlation coefficient, r value, was employed to assess the linear correlation between two variables (e.g., HCC perfusion change vs. HCC size change) (17). Data points having higher than two SDs from the regression line were considered outliers. Receiver-operation characteristic curve analysis was used to measure the sensitivity and specificity of an imaging biomarker to predict the progressive disease, when the cut-point was set to 0% (18). The p-value of less than 0.05 was considered statistically significant in this manuscript. SAS, v9.4 (SAS Institute, Cary NC) was used for all statistical analyses.

Results

Figure 2. Perfusion change in hepatocellular carcinoma (HCC) during atezolizumab plus bevacizumab treatment. (a) Representative HCC before (top row) and at 9±2 weeks after starting atezolizumab plus bevacizumab treatment (bottom row), categorized as partial response (left column), stable disease (middle column), and progressive disease (right column) based on mRECIST. HCC perfusion is shown on a color scale, estimated with the tumor-to-liver signal ratio in the arterial phase. (b, c) The change of (b) HCC perfusion (tumor-to-liver signal ratio in the arterial phase) and (c) HCC size for 9±2 weeks of atezolizumab plus bevacizumab treatment in the complete or partial response (CR/PR) group, stable disease (SD) group, and progressive disease (PD) group, when the therapeutic effect was determined using mRECIST at 21±5 weeks after starting treatment.

Three subjects had complete responses (CR), five subjects had partial responses (PR), eight subjects had stable diseases (SD), and three subjects had progressive diseases (PD). The mean baseline tumor long axis was 59±47 mm. Figure 2a shows the representative tumors of the PR, SD, and PD groups in the baseline and the first follow-up scans. The tumor regions are presented in a color scale, while the color represents the tumor-to-liver signal ratio (that is, perfusion). Figure 2b shows the tumor perfusion changes of the favorably responding (CR/PR) group (-26±18%), the SD group (-24±12%), and the PD group (9±13%) for nine weeks of treatment with atezolizumab with or without bevacizumab. There was no difference in perfusion changes between the CR/PR and SD groups (p=0.7368), but a significant difference was found between the CR/PR and PD groups (p=0.0135) or between the SD and PD groups (p=0.0040). Figure 2c shows the tumor-size changes of the CR/PR group (-38±15%), the SD group (-1±11%), and the PD group (17±10%) during the same period of treatment. The tumor-size change in the CR/PR group was significantly different from that in the SD group (p<0.0001) or the PD group (p=0.0003). The tumor-size change in the SD group also was different from that of the PD group (p=0.0285).

Without three atezolizumab monotherapy cases, the tumor perfusion changes for nine weeks in the CR/PR, SD, and PD groups were − 29 ± 17% (n = 7), -23 ± 13% (n = 7), and 3 ± 12% (n = 2), respectively, while the tumor size changes were − 38 ± 16%, -3 ± 11%, and 15 ± 13%, respectively. The perfusion changes in the three tumors treated with atezolizumab alone were − 4% (PR), -28% (SD), and 19% (PD), and the tumor size changes were − 34% (PR), 7% (SD), and 22% (PD).

The tumor perfusion change during nine weeks of treatment using atezolizumab with/without bevacizumab yielded 100% sensitivity and 94% specificity to predict the progressive diseases at 21 weeks after therapy initiation. The tumor size change during the same period also yielded 100% sensitivity, but lower specificity (69%). When the data of atezolizumab monotherapy were excluded, the tumor perfusion change yielded 100% sensitivity and 93% specificity, and the tumor size change yielded 100% sensitivity and 71% specificity.

Figure 3a shows a significant correlation between the tumor perfusion change for nine weeks and the tumor size change for 21 weeks (p = 0.0004, r = 0.81) when the CR group and two outliers in the PR group indicated with the circles are excluded. If the two outliers in the PR group are included, the p and r values are 0.0305 and 0.54, respectively. When the data retrieved only from the CT images were used (n = 9), the tumor perfusion change for nine weeks were significantly correlated with the tumor size change for 21 weeks. However, the correlation was not significant (p = 0.2415, r = 0.51) when the data retrieved only from the MRI images were used (n = 7). Figure 3b shows a significant correlation between the tumor size change for nine weeks and the tumor size change for 21 weeks (p < 0.0001, r = 0.92) when the CR group is excluded.

Discussion

To our knowledge, this is the first report that the early HCC perfusion change can predict the long-term therapeutic efficacy of atezolizumab plus bevacizumab in human subjects. All FDA-approved tyrosine kinase inhibitors for HCC treatment have an anti-angiogenic effect. Immune-checkpoint inhibitors, however, do not directly induce an anti-angiogenic effect, so it was unknown how the HCC perfusion would be changed. A recent preclinical study suggested that an effective immune checkpoint inhibitor would increase the HCC perfusion (13), and it might be further increased when combined with bevacizumab due to the vascular normalization effect (14). In this study, however, we demonstrated the perfusion of HCC favorably responding to the atezolizumab plus bevacizumab was significantly decreased in human subjects. We presume this discrepancy stemmed from the difference in the follow-up timing. Zheng et al monitored the perfusion change only for 1.5 weeks after therapy initiation (13), and Shigeta et al analyzed the microvessel density only two weeks after therapy initiation (14). In contrast, the first follow-up scan of human subjects was approximately nine weeks after starting the therapy. Therefore, the HCC perfusion may increase during the early treatment with atezolizumab plus bevacizumab, but the vessel density may decrease thereafter as the tumor cell density is reduced.

In this study, the tumor perfusion change for nine weeks did not outperform the tumor size change during the same period in predicting the long-term HCC response to therapy. However, as the physiologic change precedes the anatomic change, we expect the perfusion change to serve as an earlier surrogate imaging biomarker if assessed accurately. We estimated the tumor perfusion with the tumor-to-liver signal ratio at the arterial phase. The CT pixel value is proportional to the contrast-agent concentration, associated with the blood perfusion. However, the contrast-agent concentration is subject to change according to the cardiac output during the image acquisition, which may lead to measurement errors. The MRI pixel value is nonlinearly proportional to the contrast-agent concentration (19), and the perfusion estimated in the MRI images could be underestimated, particularly for the hyper-fused tumors. Besides, the local flip angles of the tumor and liver regions could be different due to B1 field non-homogeneity in MRI, inducing additional measurement errors (20). This may explain that the CT-based perfusion data were significantly correlated with the long-term tumor-size change, but the MRI-based perfusion data were not. The volume transfer constant, Ktrans, could be retrieved from dynamic contrast-enhanced (DCE) CT or DCE-MRI to measure the tissue perfusion accurately (21). DCE-CT allows reproducible Ktrans measurement, but the high radiation dose remains a concern (22). On the other hand, DCE-MRI does not use ionizing radiation, but poor data reproducibility remains a concern (23). We developed a perfusion phantom that significantly improves the reproducibility of abdominal tissue perfusion measurement from DCE-MRI (24). The HCC perfusion may be more accurately measured using quantitative DCE-MRI after the phantom-based error correction, serving as a reliable surrogate imaging biomarker of HCC response to atezolizumab plus bevacizumab.

The earliest time point of the HCC follow-up scan in routine clinical practice would be at six weeks after therapy initiation due to the reimbursement policies of most health insurance companies. We expect perfusion imaging to assist the clinical decision, especially when the tumor response is indeterministic based on mRECIST. Even if the early perfusion change predicts the long-term tumor size change, not all anatomic changes in tumors will lead to the improvement of clinical outcomes. Therefore, the imaging data, including size and perfusion, would need to be interpreted with clinical and demographic information (e.g., race/ethnicity, age, sex, etc.) to make more appropriate clinical decisions.

In conclusion, the perfusion change of HCC may serve as an early surrogate imaging biomarker of atezolizumab plus bevacizumab. The study results will need to be validated in a larger population, ideally in a multi-institutional setting.

Declarations

ACKNOWLEDGMENT

The authors thank Mr. Morgan Amos, Ms. Amanda Richardson, Ms. Haley Hendrix, and Ms. Brandi Barger for collecting clinical data. This study was supported by the department of radiology incentive grant at UAB and the UAB comprehensive cancer center grant, P30 CA13148.

Funding

This work was supported by the department of radiology incentive grant at UAB and the UAB comprehensive cancer center grant, P30 CA13148.

Competing interests

The authors have no relevant financial or non-financial interests to disclose.

Author contributions

All authors contributed to the study conception and design. Data collection and analysis were performed by Ezinwanne Onuoha and Andrew Smith. The first draft of the manuscript was written by Ezinwanne Onuoha, and all authors commented on previous versions of the manuscript. All authors read and approved the final manuscript.

Data Availability

All data were collected from UAB clinic data base, which are not publicly available.

Ethics approval

This study was performed in line with the principles of the Declaration of Helsinki. 

Consent to participate

Informed consent was exempted for this retrospective study from UAB IRB.

Consent to publish

Informed consent was exempted for this retrospective study from UAB IRB.

References

  1. Mittal S, El-Serag HB. Epidemiology of hepatocellular carcinoma: consider the population. J Clin Gastroenterol. 2013;47 Suppl:S2-6.
  2. Park JW, Chen M, Colombo M, et al. Global patterns of hepatocellular carcinoma management from diagnosis to death: the BRIDGE Study. Liver Int. 2015;35(9):2155-66.
  3. Yang Y. Cancer immunotherapy: harnessing the immune system to battle cancer. J Clin Invest. 2015;125(9):3335-7.
  4. Pardoll DM. The blockade of immune checkpoints in cancer immunotherapy. Nat Rev Cancer. 2012;12(4):252-64.
  5. Finn RS, Qin S, Ikeda M, et al. Atezolizumab plus Bevacizumab in Unresectable Hepatocellular Carcinoma. N Engl J Med. 2020;382(20):1894-905.
  6. Yoo C, Kim JH, Ryu MH, et al. Clinical Outcomes with Multikinase Inhibitors after Progression on First-Line Atezolizumab plus Bevacizumab in Patients with Advanced Hepatocellular Carcinoma: A Multinational Multicenter Retrospective Study. Liver Cancer. 2021;10(2):107-14.
  7. Andrews A. Treating with Checkpoint Inhibitors-Figure $1 Million per Patient. Am Health Drug Benefits. 2015;8(Spec Issue):9.
  8. Verma V, Sprave T, Haque W, et al. A systematic review of the cost and cost-effectiveness studies of immune checkpoint inhibitors. J Immunother Cancer. 2018;6(1):128.
  9. Wan X, Luo X, Tan C, Zeng X, Zhang Y, Peng L. First-line atezolizumab in addition to bevacizumab plus chemotherapy for metastatic, nonsquamous non-small cell lung cancer: A United States-based cost-effectiveness analysis. Cancer. 2019;125(20):3526-34.
  10. Darvin P, Toor SM, Sasidharan Nair V, Elkord E. Immune checkpoint inhibitors: recent progress and potential biomarkers. Exp Mol Med. 2018;50(12):1-11.
  11. Liu X, Qin S. Immune Checkpoint Inhibitors in Hepatocellular Carcinoma: Opportunities and Challenges. Oncologist. 2019;24(Suppl 1):S3-S10.
  12. Lencioni R, Llovet JM. Modified RECIST (mRECIST) assessment for hepatocellular carcinoma. Semin Liver Dis. 2010;30(1):52-60.
  13. Zheng X, Fang Z, Liu X, et al. Increased vessel perfusion predicts the efficacy of immune checkpoint blockade. J Clin Invest. 2018;128(5):2104-15.
  14. Shigeta K, Datta M, Hato T, et al. Dual Programmed Death Receptor-1 and Vascular Endothelial Growth Factor Receptor-2 Blockade Promotes Vascular Normalization and Enhances Antitumor Immune Responses in Hepatocellular Carcinoma. Hepatology. 2020;71(4):1247-61.
  15. Mpekris F, Voutouri C, Baish JW, et al. Combining microenvironment normalization strategies to improve cancer immunotherapy. Proc Natl Acad Sci U S A. 2020;117(7):3728-37.
  16. Neter J, Kutner MH, Nachtsheim JC, Wasserman W. Applied linear statistical models. Fourth ed. Columbus: The McGraw-Hill Companies, Inc., 1996.
  17. Rodgers JL, Nicewander WA. Thirteen ways to look at the correlation coefficient. The American Statistician. 1988;42(1):59-66.
  18. Fawcett T. An introduction to ROC analysis. Pattern Recognition Letters. 2006;27(8):861-74.
  19. Heilmann M, Kiessling F, Enderlin M, Schad LR. Determination of pharmacokinetic parameters in DCE MRI: Consequence of nonlinearity between contrast agent concentration and signal intensity. Invest Radiol. 2006;41(6):536-43.
  20. Gabriel C, Gabriel S, Corthout E. The dielectric properties of biological tissues: I. Literature survey. Phys Med Biol. 1996;41(11):2231-49.
  21. Ingrisch M, Sourbron S. Tracer-kinetic modeling of dynamic contrast-enhanced MRI and CT: a primer. J Pharmacokinet Pharmacodyn. 2013;40(3):281-300.
  22. Cao SE, Zhang LQ, Kuang SC, et al. Multiphase convolutional dense network for the classification of focal liver lesions on dynamic contrast-enhanced computed tomography. World J Gastroenterol. 2020;26(25):3660-72.
  23. Kim H. Variability in Quantitative DCE-MRI: Sources and Solutions. J Nat Sci. 2018;4(1).
  24. Holland MD, Morales A, Simmons S, et al. Disposable point-of-care portable perfusion phantom for quantitative DCE-MRI. Med Phys. 2022;49(1):271-81.

Tables

Table 1. Mechanism of current therapeutic regimens for advanced hepatocellular carcinoma (HCC) patients together with the FDA approval years. PD-L1: Programmed death –ligand 1; VGEF: vascular growth endothelial factor; CTLA-4: cytotoxic T-lymphocyte associated protein 4; PD-1: Programmed death 1; VEGFR2: vascular endothelial growth factor receptor 2.

Agents

Mechanism of Action

Year

Atezolizumab + Bevacizumab

PD-L1 inhibitor + VGEF inhibitor

2020

Durvalumab + Tremelimumab

PD-L1 inhibitor + CTLA-4 inhibitor

2020

Nivolumab + Ipilmumab

PD-1 inhibitor + CTLA-4 inhibitor

2020

Nivolumab

PD-1 inhibitor

2020

Pembrolizumab

PD-1 inhibitor

2018

Ramucirumab

VEGFR2 inhibitor

2019

Cabozantinib

Tyrosine Kinase Inhibitor

2019

Regorafenib

Tyrosine kinase inhibitor

2018

Lenvatinib

Tyrosine Kinase inhibitor

2018

Sorafenib

Tyrosine Kinase inhibitor

2007