Yttrium-90 hepatic radioembolization: clinical review and current techniques in interventional radiology and personalized dosimetry

Abstract

In recent years, yttrium-90 (⁹⁰Y) microsphere radioembolization has been establishing itself as a safe and efficacious treatment for both primary and metastatic liver cancers. This extends to both first-line therapies as well as in the salvage setting. In addition, radioembolization appears efficacious for patients with portal vein thrombosis, which is currently a contraindication for surgery, transplantation and transarterial chemoembolization. This article reviews the efficacy and expanding use of ⁹⁰Y microsphere radioembolization with an added emphasis on recent advances in personalized dosimetry and interventional radiology techniques. Directions for future research into combination therapies with radioembolization and expansion into sites other than the liver are also explored.

INTRODUCTION

Yttrium-90 (⁹⁰Y) microsphere radioembolization is a promising treatment modality that has emerged for the management of patients with liver cancer. Liver tumours, both primary and metastatic, form a large proportion of solid tumours with a variety of therapeutic options. Radioembolization with the radioisotope ⁹⁰Y is a procedure in which resin or glass microspheres containing ⁹⁰Y are administered directly into the hepatic arteries which supply the hepatic tumour.¹

Historical background

One of the first pioneers in performing radioembolization with ⁹⁰Y microspheres for colorectal cancer (CRC) hepatic metastases was Ariel,² a surgeon from New York, in animal experiments and subsequently in two human volunteers. Nolan and Grady³ then described the use of ⁹⁰Y oxide to treat hepatocellular carcinoma (HCC) in humans. Subsequent clinical trials involved several successful treatments of metastatic CRC to the liver.^4,5 This therapeutic modality has since been incorporated into early nuclear medicine literature following its initial clinical applications.^{6–8 90}Y microsphere radioembolization is presently used for local control of both primary as well as metastatic liver tumours which are unresectable.

Radioembolization concepts

Unlike the majority of organs, there exists a dual hepatic blood supply for the liver: the portal vein and hepatic artery. Primary HCC nodules demonstrate reduction in portal vein and normal hepatic artery supply, whereas abnormal artery supply (intranodular arterial supply through newly formed abnormal arteries) gradually increases.⁹ Metastatic liver tumours >3 mm derive 80–100% of their blood supply from arterial rather than portal hepatic circulation.¹⁰ These concepts are the basis for intra-arterial administration of ⁹⁰Y microspheres. Two parts are present in this radioembolization procedure: embolization and brachytherapy. The angiographic end points of embolization and sometimes stasis as well as the requirement to adjust the delivery according to angiographic findings under fluoroscopy define this therapy as an embolization procedure. The administration of radiation with dosimetry based on tumour and target volumes define this therapy as a brachytherapy procedure.¹¹ Radioembolization is thus a combination of both radiation with embolization although the tissue effects are primarily mediated by radiation injury. Although the terms “selective internal radiation therapy” and “radioembolization” have been used synonymously, the Society of Interventional Radiology (Fairfax, VA) prefers the term “radioembolization”.¹²

Yttrium-90 properties and microspheres

⁹⁰Y is a beta emitter with a 64.2-h physical half-life, in which up to 94% of the ⁹⁰Y microspheres radiation dose can be delivered during the first 11 days following treatment, after which it decays into stable zirconium. ⁹⁰Y microspheres entrapped in the microvasculature of the liver emit β-radiation with a mean energy of 0.94 MeV, maximum energy of 2.26 MeV, mean tissue penetration of 2.5 mm and maximum penetration of 11 mm. By Medical Internal Radiation Dose (MIRD) principle, one gigabecquerel (GBq) of ⁹⁰Y uniformly distributed throughout 1 kg of tissue provides an absorbed dose of approximately 50 Gy.¹¹

There are currently two types of ⁹⁰Y microspheres available commercially. TheraSphere™ (glass microspheres; BTG, London, UK) was approved by the US Food and Drug Administration in 1999 for the therapy of unresectable HCC in patients.¹ SIR-Spheres™ (resin microspheres; Sirtex Medical Limited, NSW, Australia) were approved by the Food and Drug Administration in 2002 for the therapy of colorectal metastases in conjunction with intrahepatic floxuridine.¹

The average size of resin microspheres is up to 30% larger than glass microspheres, whereas specific activity is higher for the glass microspheres. This translates to lesser particles infused per GBq of ⁹⁰Y activity of TheraSphere than SIR-Spheres. Glass microspheres contain 2500 Bq per microsphere and approximately 1–2 million microspheres are infused for a typical patient. Resin microspheres contain approximately 50 Bq per microsphere, and a typical treatment contains 40–60 million microspheres. This may have an embolic effect resulting in slowed antegrade hepatic arterial flow. Table 1 shows a comparison of the properties of the glass and resin microspheres.

Table 1.

Properties of glass and resin yttrium-90 microspheres

Characteristics	Glass microspheres	Resin microspheres
Trade name	TheraSphere™	SIR-Spheres™
Diameter of microsphere (µm)	20–30	20–60
Specific gravity (g dl−1)	3.6	1.6
Specific activity (Bq per sphere)	2500	50
Number of microspheres per 3-GBq vial	1.2 × 106	40–80 × 106
Material	Glass with yttrium in matrix	Resin with bound yttrium

TheraSphere™ obtained from BTG, London, UK and SIR-Spheres™ from Sirtex Medical Limited, NSW, Australia.

Resin microspheres are received as 3-GBq vials that can be dispensed according to the prescribed activity for each patient. Glass microspheres are generally planned for arrival a few days before the treatment, and based on the decay rate, the entire vial is delivered to the tumour.

Yttrium-90 DOSE ACTIVITY PRESCRIPTION

For the resin microspheres, there are generally two accepted methods for calculating the prescribed activity: body surface area (BSA) method and the partition model. The BSA method is the most commonly used but is semi-empirical. The second method is the “partition model”, which enables personalized radiation planning based on absorbed doses (Gy). A third method which is the earliest and empirical method is no longer recommended.

Limitations of semi-empirical methods

Experts in radionuclide internal dosimetry have long criticised the common practice of empiric activity prescription for therapeutic purposes. Even the common practice of modifying prescribed activities based on body weight or BSA is semi-empirical and stems historically from the experiences of systemic chemotherapy. However, therapeutic radiopharmaceuticals exert their biological response primarily by the deterministic effects of radiation and are therefore not a “drug” and should not be regarded as such. Implicit within empirical or semi-empirical methods are biophysical assumptions which the user has knowingly or unknowingly accepted to be applicable to their patient. Any studies demonstrating apparent safety in spite of empirical or semi-empirical methods only lead to the logical conclusion that these methods have a conservative tendency to underdose, sacrificing efficacy for safety.

In the context of resin microsphere activity prescription, the BSA method¹³ is the most commonly used method owing to its simplicity, attributable to its many assumptions.¹⁴ The BSA method was devised more than a decade ago¹⁵ when there was little emphasis on the importance of microsphere radiobiology and before the advent of hybrid scintigraphy. The BSA method was therefore clinically appropriate within the context of its initial design. However, its semi-empirical nature inherently limits its ability to scientifically evolve with new and better technology. In the face of widespread availability of hybrid tomographic scintigraphy today, the ethics of subjecting patients to a decade-old treatment paradigm is questionable.

Recently, Bernardini et al compared dosimetric parameters between BSA methodology vs partition model on 28 patients of various tumour types. From their heterogeneous patient cohort, they found no correlation between the activities prescribed by BSA methodology and tumour-to-normal liver (T/N) ratios or percentage tumour involvement. On the other hand, activities prescribed by the partition model were shown to correlate with T/N ratios and percentage tumour involvement.¹⁶ Another recent study by Lam et al¹⁷ involving 45 patients with colorectal liver metastasis did not show correlation between activities prescribed by the BSA method and absorbed doses averaged across the whole liver. All of these observations are not surprising given the semi-empirical nature of the BSA method because its formularism sacrifices personalization in favour of simplicity.¹⁴

Personalized predictive dosimetry

The modern era of personalized medicine demands the measurement of patient-specific biophysical parameters to individualize treatment so as to maximize the desired effect while minimizing toxicity. This is especially important in oncology because patients may be frail or have limited prognosis. Hybrid tomographic scintigraphy with integrated CT represents a technological leap from planar scintigraphy and has the potential to modernize radionuclide therapy from an art into a science to achieve predictable and reproducible results. Single-photon emission tomography with integrated CT (SPECT/CT) and positron emission tomography with integrated CT (PET/CT) overcomes many of the technological limitations which have traditionally plagued planar and single-photon emission tomography (SPECT)-only scintigraphy to allow more accurate measurements of dosimetric parameters.

⁹⁰Y microsphere radioembolization is well placed to lead the science in personalized radionuclide therapy because the permanently implanted microspheres negate tedious time–activity measurements. Provided the angiographic conditions, especially catheter position, between exploratory angiography and radioembolization are identical, the technetium-99m macroaggregated albumin (^99mTc MAA) SPECT/CT may be regarded as a “radiation simulation” study^18,19—a concept demonstrated for both resin¹⁹ and glass²⁰ microspheres. Today, the simplest method of personalized predictive dosimetry for radioembolization is by MIRD macrodosimetry.

MIRD macrodosimetry has been applied to resin microspheres as early as the 1990s, also known as the “partition model”.²¹ The MIRD equations are widely available from the literature.^18,21,22 The partition model mathematically describes the biodistribution of microspheres in a typical patient into three dosimetric compartments—tumour, non-tumorous liver and the lung. The use of count ratios from the ^99mTc MAA mapping study and derivative parameters such as the hepatopulmonary shunt fraction and the T/N ratio allow the desired absorbed doses and activities to be estimated in each of the three compartments. This allows predictive dosimetry to be performed for the tumour, non-tumorous liver and the lung prior to radioembolization. With improvements in angiographic methods over the years, the original partition model has now evolved to take into account selective or superselective radioembolization depending on the case-specific clinical, angiographic and dosimetric circumstances.¹⁹ This has developed into a “multipartition” model where vascular territories are mapped by catheter-directed CT [also known as intra-arterial CT (IACT) or CT hepatic angiography] and the ^99mTc MAA biodistribution is tomographically mapped by SPECT/CT.¹⁹ In addition, the multipartition model improves radiation planning for large tumours that are supplied by multiple arteries by measuring the different T/N ratios for each arterial territory, allowing treatment personalization and absorbed dose optimization.¹⁹ For glass microspheres, its manufacturer advocates a MIRD-based dosimetry where the intended absorbed dose is averaged across the entire target arterial territory, and consequently, the absorbed doses to tumour and non-tumorous liver are not separately determined.¹⁸ However, the same general principles of MIRD macrodosimetry for tumour, non-tumorous liver and the lung also apply to glass microspheres.²⁰ Our recommendations for safe, effective and personalized predictive dosimetry by MIRD macrodosimetry for radioembolization are summarized in Table 2.

Table 2.

Recommendations for radioembolization personalized predictive dosimetry

Understand	1. Clinical history and baseline clinical status
	2. Angiographic findings and technical challenges
	3. Catheter positions for 99mTc MAA injections
Measure	1. Hepatopulmonary shunt fraction by 99mTc MAA SPECT/CT23
	2. Patient-specific lung mass by CT densitovolumetry23
	3. Tumour and normal tissue volumes by catheter-directed CT angiography19
	4. T/N ratios for each target arterial territory by 99mTc MAA SPECT/CT19
Decide	1. The dose limiting organ
	2. Whether the MIRD assumption of uniform activity distribution is reasonable
	3. Extent of dosimetric uncertainty and size of safety margin
Formulate	1. Apply MIRD macrodosimetry based on dose limiting organ
	2. Derive the safest achievable tumour absorbed dose
	3. Adjust radiation plan based on extent of dosimetric uncertainty
	4. Prescribe 90Y activity in accordance to radiation plan
Communicate	1. Discuss radiation plan with patient and multidisciplinary team
	2. Agree on treatment intent based on safest achievable tumour absorbed dose
	3. Ensure catheter positions for radioembolization are identical to 99mTc MAA injections

^99mTc, technetium-99m; MAA, macroaggregated albumin; MIRD, medical internal radiation dose; SPECT, single-photon emission tomography; T/N, tumour-to-normal liver; ⁹⁰Y, yttrium-90.

Dosimetric uncertainty

Critics of predictive dosimetry based on ^99mTc MAA SPECT/CT often cite studies which apparently do not show correlation between ^99mTc MAA biodistribution and clinical outcomes. However, such studies were often confounded by a lack of radiobiological perspective and mislead less experienced readers.^24–26 Others challenge ^99mTc MAA predictive dosimetry based on the physical differences between MAA vs resin or glass microspheres and differences in injected microparticle load. Although these physical differences are undeniably true, the concept often overlooked is that ^99mTc MAA SPECT/CT is meant to depict what the microsphere biodistribution is likely to be and was never intended to exactly replicate the actual microsphere biodistribution. When formulating the radiation plan, the nuclear medicine physician also considers information from angiography and relevant diagnostic imaging and clinical data to account for real-life dosimetric uncertainties. The extent of uncertainty is reflected in the radiation plan by how large or small the absorbed dose safety margin should be. This is then balanced with the safest achievable tumour absorbed dose and achievable treatment intent. It follows that some radioembolization candidates may be disqualified on the basis of unfavourable dosimetric parameters which risk unsafe or futile therapy. Preliminary data now exist to describe the general statistical limits of dosimetric uncertainty across a population of patients undergoing radioembolization using resin microspheres.²⁶

The key requirement for reliable predictive dosimetry is to obtain accurate input dosimetric parameters, of which tissue masses and count ratios are the most important. The increasingly common practice of lobar or segmental radioembolization demands more accuracy in tissue mass measurements within targeted arterial territories. This may be achieved by catheter-directed CT angiography, either by transferring the patient between the angiography suite and CT scanner or by hybrid angio-CT systems where available.¹⁹ Today, SPECT/CT is widely available for measurement of volumetric count ratios. The clinical use of ^99mTc MAA SPECT/CT of the abdomen for calculating artery-specific T/N ratios, guided by catheter-directed CT, has been described.¹⁹ By including the lungs into the ^99mTc MAA SPECT/CT acquisition, attenuation-corrected hepatopulmonary shunt fractions and patient-specific lung mass estimates by CT densitovolumetry can also be obtained.²³

Although MIRD macrodosimetry is conceptually simple to apply, its main weakness is its assumption of uniform activity distribution. Microspheres are always heterogeneously distributed, hence MIRD macrodosimetry should be used cautiously in cases where the activity biodistribution is highly heterogeneous. Such situations may be encountered in very large tumours or tumours that have been partially treated. Activity heterogeneity may be visualized directly by reviewing the ^99mTc MAA SPECT/CT images or graphically expressed as dose–volume histograms. This is analogous to the external beam radiation therapy concept of tomographic radiation planning using isodose maps and dose–volume histograms. Predictive dosimetry based on ^99mTc MAA SPECT/CT dose–volume histograms is still the subject of research but promises to optimize radiation planning by overcoming the MIRD assumption of uniform activity distribution.²⁷

Absorbed dose thresholds

Although the principles of MIRD macrodosimetry may be applied to resin and glass microspheres, the biophysical characteristics of each device is different and therefore exert different tissue radiobiology. In other words, dose–response data for resin vs glass microspheres are not directly interchangeable unless additional calculations, such as the biologically effective dose (BED), have been performed.¹⁸ It follows that institutions where resin and glass microspheres are both available should separate these two devices when analysing their data.

Based on an expert panel, HCC mean absorbed dose thresholds for resin microspheres are >120 Gy for tumour response, <50 Gy to non-tumorous whole liver and <20 Gy to the lung.²⁸ Extrapolating from external beam radiation therapy data, resin microsphere normal tissue complication probability for non-tumorous whole liver was estimated by Cremonesi et al¹⁸ to be TD_5/5 35 Gy and TD_50/5 44 Gy, and for the lung to be TD_5/5 23 Gy and TD_50/5 30 Gy, where TD_5/5 and TD_50/5 represent the tolerance doses (TDs) that would result in 5% and 50% risk of severe complications within 5 years, respectively. Dose–volume histogram analysis of ⁹⁰Y PET/CT by Kao et al²⁹ identified the threshold for HCC complete response to be D₇₀ >100 Gy using resin microspheres, where D₇₀ is the minimum absorbed dose delivered to 70% tumour volume.

For colorectal liver metastasis treated with resin microspheres, Lam et al³⁰ evaluated a heterogeneous cohort of 25 patients and found responders to have a mean tumour absorbed dose of 82.7 ± 23.9 Gy, whereas non-responders had 31.0 ± 10.9 Gy. Flamen et al³¹ assessed the change in total lesion glycolysis of 39 lesions of colorectal liver metastasis and found responding lesions to have a median tumour absorbed dose of 66 Gy (95% CI 32–159 Gy), whereas lesions with poor response had 29 Gy (95% CI 1–98 Gy). An expert panel recommended the mean absorbed dose threshold for non-tumorous whole liver to be 50–70 Gy for colorectal liver metastasis, depending on liver reserve and prior systemic therapy.²⁸

For glass microspheres using Day 3 post-calibration specific activity of 1250 Bq/sphere delivered in sequential lobar radioembolization, Garin et al²⁰ found HCC thresholds to be >205 Gy for tumour response, <120 Gy to healthy liver and <30 Gy to the lung. Chiesa et al³² used glass microspheres 3.75 days post-calibration for unilobar radioembolization of HCC and found TCP₅₀ to be approximately 500 Gy depending on the analysis method, where TCP₅₀ is the tumour control probability representing 50% probability of lesion response. TCP₅₀ was highly dependent on the lesion size: 250 Gy for lesions <10 cm³ and >1000 Gy for larger lesions. Normal tissue complication probability for non-tumorous whole liver was found to be TD_15/0.5 75 Gy, where TD_15/0.5 is the TD that would result in 15% risk of severe complications within 0.5 years (i.e. 6 months) for Child-Pugh A patients without complete portal vein obstruction.³² There is currently no data on glass microspheres tumour absorbed dose thresholds for colorectal liver metastasis. It should be cautioned that “extended shelf-life” glass microspheres are likely to have different radiobiological characteristics due to lower specific activity and therefore the above-mentioned thresholds may not be wholly applicable.

For repeated radioembolization to the same arterial territory, the cumulative absorbed dose to healthy tissue and time interval between radioembolizations are important safety considerations. However, radiobiological guidance on this issue is scarce in the literature for both resin and glass microspheres and further research is warranted. All of the above data are summarized in Table 3.

Table 3.

Absorbed dose thresholds (Gy) for yttrium-90 radioembolization

Tumour/normal organ	SIR-Spheres™ (Gy)	TheraSphere™ (Gy)
HCC19,20,28,29,32	Mean 91–120 (PR) D70 >100 (CR)	Mean 205 (PR) TCP50 500 (PR)
Colorectal30,31	Mean 66–83 (PR)	No data
Non-tumorous whole liver18,20,28,32	Mean 50–70	Mean 120
	TD5/5 35	TD15/0.5 75
	TD50/5 44
Lung18,20,28	Mean 20	Mean 30
	TD5/5 23
	TD50/5 30

CR, complete lesion response; D₇₀, minimum absorbed dose delivered to 70% tumour volume; HCC, hepatocellular carcinoma; PR, at least partial lesion response; TCP₅₀, tumour control probability representing 50% probability of lesion response; TD_5/5 and TD_50/5, tolerance doses that would result in 5% and 50% risk of severe complications within 5 years.

TheraSphere™ obtained from BTG, London, UK and SIR-Spheres™ from Sirtex Medical Limited, NSW, Australia.

RADIOEMBOLIZATION: CLINICAL INDICATIONS, EFFICACY AND CONSIDERATIONS

The role for radioembolization in the management of primary and secondary liver malignancies

The liver is a common organ involved in malignancies which may be primary to the liver or metastatic from elsewhere. The most common form of primary liver tumour is HCC. HCC is a complex disease and management is influenced by various clinical parameters such as tumour burden and liver function. Conceptually a systematic, protocolized approach would ensure optimal therapy and improve patient outcomes.³³

HCC continues to be the sixth most common cancer, contributing the third highest cancer mortality worldwide.³⁴ Surgical resection, transplantation or radiofrequency ablation (RFA) in selected cases, is potentially curative in early HCC. Unfortunately, the large majority of HCCs present in locally advanced or metastatic stages or are poor surgical candidates for resection.³⁵ The median survival of untreated inoperable HCC is approximately 3–8 months. Hence, there is a great impetus for the efficacious employment of locoregional therapy.

The liver is also the dominant metastatic site for patients with CRC. Pre-existing data from the 1960s and 1970s have shown that the median survival of patients who receive no treatment ranges between 3 to 12 months, with an overall median survival of 7 months.³⁶ Although surgery offers the most favourable outcomes, liver metastases resectability rates at the time of diagnosis are generally low and hence only a small proportion of patients may benefit from surgical approach. When surgery is not feasible, cytoreduction improves functionality and may prolong survival. As compared with cryoablation, RFA or microwave ablation techniques, radioembolization is a promising cytoreductive modality that encompasses the entire liver or a lobar volume. Even in a salvage setting, radioembolization appears to be a reasonable technique with a relative wide safety margin. The use of radioembolization against liver metastases from different primary cancers is also expanding. Such examples would include metastases from breast, uveal melanoma, neuroendocrine tumours, sarcoma, oesophageal, endometrial, lung, ovarian and squamous cell carcinoma of the anus.^37,38

In spite of its relatively recent development, radioembolization has yielded great promise and potential as a safe and efficacious treatment for locally advanced liver cancer.^39,40 In both salvage and first-line therapy, radioembolization has demonstrated high response rates. A number of Phase III trials have pitted radioembolization against various treatment modalities such as systemic chemotherapy and transarterial chemoembolization. Currently, the Phase III trials are still ongoing and only with their completion would the true efficacy of radioembolization be known.⁴¹

Evidence for radioembolization in primary and secondary liver malignancies

Several studies have shown the efficacy of radioembolization in liver malignancies although indications and patient selection are varied. Summarized in Table 4 are studies looking at the median survival of various liver tumours post radioembolization.⁴² From the summary, it is reasonable to conclude that radioembolization is efficacious in both primary and secondary liver tumours. Specifically, there is very high interest in the clinical use of radioembolization in locally advanced HCC because of the paucity of efficacious therapy in this group of patients.

Table 4.

Summary of outcomes of yttrium-90 microsphere radioembolization in liver malignancies

Study	Study design	Subjects, n	Median survival (months)
Primary liver tumours
Hepatocellular CA
Kulik et al 200843	Prospective	108	15.6
Sangro et al 201244	Prospective	325	13.1
Cholangiocarcinoma
Saxena et al 201045	Prospective	25	9.3
Ibrahim et al 200846	Prospective	24	14.9
Rafi et al 201347	Retrospective	19	11.5
Hoffman et al 201248	Retrospective	33	22.0
Haug et al 201149	Retrospective	26	12.8
Sarcoma
Oh et al 201350	Prospective	11	8.7
Secondary liver tumours
Metastases from colorectal CA
SIRFLOX trial51	Randomized controlled trial	518	10.7
FOXFIRE52	Randomized controlled trial	364	Ongoing
Lewandowski et al 201453	Prospective	214	10.6
Benson et al 201354	Prospective	151	8.8
Metastases from breast CA
Saxena et al 201455	Prospective	77	11.5
Jakobs et al 200856	Prospective	30	14.2
Coldwell et al 201257	Prospective	44	14.0
Cianni et al 201358	Retrospective	77	11.5
Metastases from pancreatic CA
Michl et al 201459	Retrospective	19	9
Metastases from lung CA
Murthy et al 200860	Prospective	6	2.7

CA, carcinoma.

Adapted from Khajornjiraphan et al⁴² with permission from Karger Publishers.

Reports of the efficacy of radioembolization in patients with vascular involvement in HCC have been particularly important because no other therapy has been shown to be efficacious. Portal vein tumour thrombosis (PVTT) occurs in 10–40% of HCC cases and is shown to be associated with an extremely poor prognosis.^61–63 The median survival of patients with PVTT in HCC is estimated to be 2–4 months and is a stark contrast to patients with HCC without PVTT (10–24 months).^63,64 This is not only attributed to the local and systemic effects of PVTT but also the limitation of treatment modalities. PVTT is currently contraindicated for curative surgery, transplantation and transarterial chemoembolization.^65,66 By contrast, radioembolization is safe in the setting of PVTT due to its low embolic effect and may even lead to portal vein revascularization. Table 5 lists the studies in which patients with PVTT demonstrate good response to radioembolization. Studies of ⁹⁰Y microsphere radioembolization in cases with portal vein thrombosis have demonstrated superiority of radioembolization in such cases.

Table 5.

Response and median survival post yttrium-90 microsphere radioembolization in patients with hepatocellular carcinoma with and without portal vein tumour thrombosis (PVTT)

Study	PVTT status	Number of patients	Overall survival (months)
Salem et al 201067	Child-Pugh A	116	17.2
	No PVTT	81	22.1
	PVTT	35	10.4
	Child-Pugh B	122	7.7
	No PVTT	65	14.9
	PVTT	57	5.6
Hilgard et al 201068	All patients	108	16.4
	No PVTT	75	16.4
	PVTT	33	10
Sangro et al 201169	All patients	325	12.8
	No PVTT	249	15.3

Adapted from Lau et al⁷⁰ with permission from Karcher Publishers.

The precise role of radioembolization in HCC can only be determined by prospective studies of which a number of them are ongoing. Table 6 summarizes some of the ongoing Phase II and III trials comparing radioembolization against other treatment modalities. Owing to differences in microsphere radiobiology, prospective studies comparing the clinical efficacy of SIR-Spheres vs TheraSphere are also warranted.

Table 6.

Phase II and III yttrium-90 (⁹⁰Y) microsphere radioembolization clinical trials in hepatocellular carcinoma

Trial number	Sponsor	Size	Centres	Therapy 1	Therapy 2	PO	Start	End	Status
NCT01135056	Singapore General Hospital, Singapore	360	26 (Asia–Pacific)	SirSpheres™	Sorafenib	OS	July 2010	July 2015	R
NCT01482442	Hôpitaux de Paris, Paris, France	400	1 (France)	SirSpheres	Sorafenib	OS	November 2011	March 2015	C
NCT01556490	Nordion, Ottawa, Canada	400	2 (USA)	TheraSphere™	Sorafenib	OS	March 2012	October 2016	R
Phase II NCT01686880	Jules Bordet Institute, Brussels, Belgium	50	1 (Belgium)	SirSpheres	–	Perioperative morbidity	September 2012	October 2015	R
Phase II NCT00712790	Singapore General Hospital, Singapore	35	4 (Asia-Pacific)	SirSpheres, sorafenib	–	TTR	June 2008	June 2009	C
Phase II NCT01126645	University of Magdeburg, Magdeburg, Germany	665	34 (Europe)	SirSpheres, sorafenib	RFA, sorafenib	TTR, OS	December 2010	September 2014	R
Phase II NCT00956930	Northwestern University, Evanston, IL	124	1(USA)	TheraSphere	TACE	TTP	August 2009	August 2018	R
Phase II NCT01381211	Ghent University, Ghent, Belgium	140	2 (Europe)	TheraSphere	DC-BEADS	TTP	September 2011	December 2016	R
NCT00109954	University of Pittsburgh, Pittsburgh, PA	120	1 (USA)	TheraSphere	Cisplatin, TACE	PFS	February 2005	–	A
Phase II NCT01900002	MD Anderson Cancer Center, Houston, TX	20	1 (USA)	TheraSphere, sorafenib	Sorafenib	Sorafenib and 90Y toxicity	September 2013	September 2017	R

A, active not recruiting; C, completed; DC BEADS, drug-eluting bead; OS, overall survival; PO, primary outcome; PPS, progression-free survival; RFA, radiofrequency ablation; TAFE, transarterial chemoembolization; R, recruiting; TTP, time to progression; TTR, time to response.

TheraSphere™ obtained from BTG, London, UK and SIR-Spheres™ from Sirtex Medical Limited, NSW, Australia.

Clinical indications for radioembolization

Currently, radioembolization is indicated for the treatment of both locally advanced primary and metastatic cancers with the aim of maintaining quality of life and improving survival. There are also reports of HCC being downstaged to receive potentially curative therapies.^43,44,70,71 Current indications for radioembolization in HCC are found in the following guidelines seen in Table 7: European Society of Medical Oncology,⁷² National Comprehensive Cancer Network,⁷³ National Institute for Health and Clinical Excellence^74–76 and National Cancer Centre Singapore's Comprehensive Liver Cancer Clinic.⁷⁷ The current guidelines on radioembolization tend to be very similar, all suggesting the prerequisite of good liver function, proper vascular access as well as low hepatopulmonary shunting in order to prevent complications.

Table 7.

Comparison of guidelines on the use of radioembolization

Guideline	Description
ESMO	Indicated for use in patients with prior TACE failure, excellent liver function, macrovascular invasion and the absence of extrahepatic disease.72 (Evidence: III)
NCCN	Radioembolization to be used in unresectable HCC which is not suitable for transplantation based on UNOS criteria (tumour ≤5 cm in diameter or 2–3 tumours ≤3 cm each, no macrovascular involvement, no extrahepatic disease), Child-Pugh A/B, no portal hypertension, suitable tumour location, adequate liver reserve.73 (Evidence: II A)
NICE	Indicated for patients with liver predominant disease with adequate liver function (bilirubin <34 μmol l−1 and synthesis function of liver normal) who are poor surgical candidates, not suitable for other locoregional ablative therapies.74–76
CLCC	Radioembolization to be used in locally advanced HCC ± vascular invasion in Child-Pugh A/B patients.77 (Evidence: II B)

CLCC, National Cancer Centre Singapore's Comprehensive Liver Cancer Clinic; ESMO, European Society of Medical Oncology; HCC, hepatocellular carcinoma; NCCN, National Comprehensive Cancer Network; NICE, National Institute for Health and Clinical Excellence; TACE, transarterial chemoembolization; UNOS, United Network for Organ Sharing.

All levels of evidence given are according to the Oxford Centre of Evidence Based Medicine.

Contraindications and complications of radioembolization

Safety during treatment is of utmost priority and hence it is important to select patients carefully. Therefore, it is important to note the limitations and conditions from which patients must be excluded from radioembolization (Table 8).

Table 8.

Exclusion criteria to yttrium-90 (⁹⁰Y) microsphere radioembolization

Exclusion criteria to 90Y radioembolization
Absolute contraindications
Intractable clinical ascites (in spite of optimal diuretic treatment) or any other clinical signs of liver failure on physical examination
Bleeding diathesis, not correctable by the standard forms of therapy
Severe peripheral vascular disease that would preclude arterial catheterization
Severe portal hypertension with hepatofugal flow
Relative contraindications
Complete main portal vein thrombosis
Adequate haematological, renal and hepatic function as follows:
Haematological
Leukocytes <2500 μl−1
Platelets <80,000 μl−1
Haemoglobin <9.5 g dl−1
Total bilirubin >2.0 mg dl−1
INR >2.0
Hepatic function
ALP >5 × institutional ULN
AST and ALT >5 × institutional ULN
Albumin <2.5 g dl−1
Renal function
Creatinine >2.0 mg dl−1
Prior hepatic external beam radiation therapy

ALP, alkaline phosphatase; ALT, alanine aminotransferase; AST, aspartate aminotransferase; INR, international normalized ratio; ULN, upper limit of normal.

SIRveNIB Trial Protocol Exclusion Criteria.⁷⁸

About one-third of patients may experience immediate short-term abdominal pain following administration of radioembolization. An increase in number of ⁹⁰Y microspheres injected is associated with this particular adverse effect. Other mild adverse effects include a slight fever for several days or lethargy and mild nausea for up to 10 days after radioembolization. These side effects are collectively known as the post-radioembolization syndrome. More serious complications will include stomach or duodenal ulcers from the reflux of ⁹⁰Y microspheres into the gastrointestinal vascular bed or radiomicrosphere hepatitis if the non-tumorous liver absorbed dose threshold is exceeded. Radiomicrosphere hepatitis can be minimized by personalized predictive dosimetry to limit the non-tumorous absorbed dose, especially for patients with pre-existing liver disease. Radiomicrosphere pneumonitis is a complication associated with significant hepatopulmonary shunting. If the ^99mTc MAA scan demonstrates significant hepatopulmonary shunting which may potentially exceed normal lung radiation tolerance, then ⁹⁰Y radioembolization either should not proceed or a lung-limiting ⁹⁰Y activity should be prescribed based on predictive dosimetry.^19,23,79

YTTRIUM-90 MICROSPHERE RADIOEMBOLIZATION: INTERVENTIONAL RADIOLOGY

Typical approach

The interventional radiologist plays a crucial role in ⁹⁰Y microsphere radioembolization and is responsible for planning the vascular approach and safe delivery of the ⁹⁰Y microspheres. In the first procedure of exploratory hepatic angiography with ^99mTc MAA simulation, it is important to delineate hepatic vascular anatomy, interrogate tumoural blood supply and seek out possible extrahepatic parasitization. The pre-procedural multidetector CT (MDCT) scan should be carefully studied. The arterial phase MDCT provides a road map of the hepatic arterial tree. Arteries likely to supply tumours should be identified along with any arterial anatomical variants.⁸⁰ The procedure typically starts with a right common femoral artery puncture to insert a vascular sheath. After which, the superior mesenteric artery is cannulated and a digital subtraction angiogram (DSA) performed to look for aberrant hepatic supply and main portal vein patency, although portal vein thrombosis is not an absolute contraindication to radioembolization.

In conventional hepatic anatomy, the celiac artery is then cannulated and a DSA performed, after which the appropriate hepatic artery is selected (usually with microcatheters and wires) and DSA performed to look for tumoural enhancement and to exclude extrahepatic supply.

Our centre's routine practice is to perform catheter-directed IACT, either with a hybrid angio-MDCT unit or a cone beam CT, from the final microcatheter position. This is to ensure adequate tumoural coverage as well as to exclude extrahepatic supply, both of which may not be readily evident on DSA. Once the interventional radiologist is satisfied with the hepatic angiography, a discussion with the attending nuclear medicine physician then takes place before the injection of the ^99mTc MAA to clarify the final approach. After the injection, all catheters, wires and the sheath are removed and appropriately disposed of in nuclear waste bags and haemostasis of the common femoral artery is secured. The patient will then be sent for a ^99mTc MAA SPECT/CT.

If the patient is suitable for treatment, radioembolization is typically performed 2 weeks later. A careful review of the ^99mTc MAA SPECT/CT and a discussion with the attending nuclear medicine physician is important, particularly if there is extrahepatic deposition of ^99mTc MAA. Our institutional preference at the Singapore General Hospital is to perform prophylactic embolization at the time of radioembolization, rather than at the time of exploratory angiography. This is to minimize unnecessary prophylactic coiling in patients who may be later disqualified on the basis of unfavourable dosimetry. Also, prophylactic coiling may induce hepatic-intestinal collaterals which may be even more unpredictable over time. On the day of radioembolization, once the necessary prophylactic coil embolizations have been completed, we will routinely perform IACT to ensure its effectiveness, i.e. to ensure no extrahepatic enhancement from the final microcatheter position. The ⁹⁰Y microspheres are then administered. Flow stasis may occur when using resin ⁹⁰Y microspheres due to a higher number of infused microspheres, which may sometimes result in incomplete delivery of the prescribed activity.^{18,81 90}Y microspheres should therefore be infused slowly. There may be a need to occasionally check for antegrade flow with a gentle contrast injection, taking care not to reflux the contrast column. At the end of the procedure, all wires, catheters and sheath are carefully removed and disposed of in nuclear waste bags. We prefer to secure haemostasis of the groin by closure device to avoid prolong contact with the puncture site in case of contamination.

Multiple injections vs vascular manipulation with coiling

The various options for radioembolization of tumours with multiple arterial supply have been well summarized by Ray et al.⁸² The strategies include splitting the prescribed activity into more than one administrative setup and using new microcatheters for each injection, flow-directed microcatheter repositioning with contrast injection and coil embolization of other hepatic arteries allowing flow redistribution to permit ⁹⁰Y microspheres injection from a single artery.⁸³ Our preference is to perform the first option: splitting the prescribed activity to different administrative setups. This allows for more predictable absorbed dose delivery as calculated by the multipartition model, as well as reduces the chance of contamination. Flow redistribution is unpredictable and introduces additional dosimetric uncertainty and is therefore the least preferred interventional technique from the perspective of radiation planning.

Intra-arterial CT

The ability to perform catheter-directed IACT, either with a hybrid angio-MDCT system or a cone beam CT has been a boon to interventional radiologists performing radioembolization. It allows us to fully appreciate tumoural and extrahepatic enhancement, thus improving patient safety. Figures 1 and 2 both show good examples of cases using IACT.

Figure 1.

Digital subtraction angiograms (DSAs) and corresponding intra-arterial CT (IACT) showing multiple arterial supplies to a single tumour. Without IACT, it may be difficult to ensure complete lesional coverage. (a, b) DSA and IACT, respectively, with the microcatheter in the right hepatic artery. (c, d) DSA and IACT, respectively, with the microcatheter in the left hepatic artery. (e, f) DSA and IACT, respectively, with the microcatheter in the right gastroepiploic artery.

Figure 2.

(a) Digital subtraction angiogram (DSA) and (b, c) intra-arterial CT (IACT) showing usefulness of IACT for detection of extrahepatic enhancement for treatment of a large hepatocellular carcinoma. IACT acquired taken with the microcatheter at the proper hepatic artery. From this position, there is a small area of gastric enhancement (c) which would otherwise be missed on DSA alone. Eventually, the activity was split and delivered into the right and left hepatic arteries without incident.

Unanticipated ^99mTc MAA deposition detected by SPECT/CT may occasionally require a second exploratory hepatic angiography with reinjection of MAA, depending on case complexity.⁸⁴ IACT can detect small areas of extrahepatic contrast enhancement which may help to reduce the need for the additional procedures. IACT also allows ⁹⁰Y microspheres to be injected into parasitic extrahepatic supply more confidently, e.g. the inferior phrenic artery.⁸⁵ From a radiation-planning perspective, IACT allows accurate assessment of target volumes to guide SPECT/CT-based personalized predictive dosimetry.^19,86

Prophylactic vessel coiling

All intrahepatic arteries that supply extrahepatic organs originating distal to the point of ⁹⁰Y microspheres release should be prophylactically embolized. At our centre, we perform the prophylactic coiling during the same session as radioembolization, following which we perform IACT to ensure no extrahepatic enhancement before injection of ⁹⁰Y microspheres.

However, it is now becoming less clear if routine prophylactic embolization of the gastroduodenal artery (GDA), cystic artery and the falciform artery are truly essential, particular with the advent of IACT.

Gastroduodenal artery

Traditionally, coiling of the GDA has been performed prior to ⁹⁰Y microspheres injection to minimize the risk of extrahepatic ⁹⁰Y microspheres deposition.¹¹ Schelhorn et al⁸⁷ demonstrated that coiling of the GDA may induce hepatic-intestinal collaterals which may in some circumstances render radioembolization unfeasible. In our practice with routine use of IACT, we have stopped routine prophylactic coiling of the GDA. If whole-liver radioembolization is required, a bilobar approach to ⁹⁰Y microsphere administration is performed.

Cystic artery

Routine prophylactic embolization of the cystic artery is not indicated. As described by Theysohn et al,⁸⁸ the majority of patients do not require cystic artery embolization. If there is significant ^99mTc MAA activity in the gallbladder wall, the final microcatheter position may either be changed to one which is distal to the origin of the cystic artery or the cystic artery may be temporarily occluded with Gelfoam^® (Pfizer Inc., New York, NY).⁸⁸

Falciform artery

A patent hepatic falciform artery (HFA) is an uncommon finding on DSA, although they are far more readily detected on IACT.⁸⁹ Burgmans et al⁸⁹ also demonstrated that the risk of symptomatic extrahepatic deposition of ⁹⁰Y microspheres into the HFA territory is rare even with a patent HFA, and thus routine embolization of the HFA is also not indicated. If there is indeed deposition of ^99mTc MAA in the HFA vascular territory, this vessel can be embolized with gelfoam or coils, and additionally, ice packs may be placed on the anterior abdomen during radioembolization.⁹⁰

CONFIRMATION OF TECHNICAL SUCCESS

Owing to biophysical differences between soluble contrast molecules vs ⁹⁰Y microspheres, technical success should not be inferred on the basis of angiographic findings alone. To confirm technical success, the biodistribution of ⁹⁰Y microspheres should be imaged by either bremsstrahlung SPECT/CT or ⁹⁰Y PET/CT. Although bremsstrahlung SPECT/CT is a significant technological improvement from planar scintigraphy, the inherent problem of poor spatial resolution still remains. It was recently shown that minuscule positron emission from ⁹⁰Y decay may be imaged by conventional PET scanners without hardware modification.^{91 90}Y PET obtains high-resolution images for improved assessment of target⁹² and non-target activity,⁹³ superior to ⁹⁰Y bremsstrahlung SPECT/CT. The higher resolution of ⁹⁰Y PET improves diagnostic confidence for technical success confirmation, radiation plan verification and non-target activity detection.^92,93 Diagnostic reporting guidelines for ⁹⁰Y PET/CT have recently been proposed.^{92 90}Y PET voxel dosimetry may also be performed if clinically indicated to obtain absorbed doses and dose–volume histograms.²⁹ There is ongoing research to address the problem of high background noise in the reconstructed images due to the unfavourable combination of low ⁹⁰Y positron emission, intrinsic radioactivity from lutetium-based PET crystals and random events.

Post-procedure imaging

As with other forms of embolotherapy, size criteria [Response Evaluation Criteria in Solid Tumours (RECIST) or World Health Organization] alone has shown to be less accurate in gauging response.⁹⁴ Indeed, tumours that responded to radioembolization may show an initial transient increase in size. In their radiologic-pathological correlation study of HCCs in explanted liver, Riaz et al⁹⁵ showed that the European Association for the Study of the Liver disease necrosis criteria most accurately predicts complete pathological necrosis. Necrosis on imaging also appears to correlate with survival in a study of 21 patients with intrahepatic cholangiocarcinoma who undergo radioembolization, both modified RECIST and European Association for the Study of the Liver criteria, when applied to the delayed phase predicted overall survival but RECIST did not.⁹⁶ For hepatic colorectal metastases, fluorine-18 fludeoxyglucose PET/CT within 6–8 weeks may predict progression-free survival (PFS).⁹⁷

FUTURE DIRECTIONS AND RESEARCH

Radioembolization to sites other than the liver

As radioembolization is essentially a form of brachytherapy, there is no logical reason why radioactive microspheres cannot be used to treat tumours located elsewhere other than the liver. In principle, radioembolization may be performed to any site in the body, provided there is accessible vasculature, favourable tumour-vs-normal tissue targeting and adequate treatment simulation to ensure safety. A general theory of predictive dosimetry based on MIRD macrodosimetry to guide radioembolization to sites other than the liver has recently been proposed.⁹⁸ Since 2013, there have been reports of radioembolization to the kidney, spleen, lung and adrenal gland.⁹⁸ A clinical trial investigating kidney radioembolization for renal cell carcinoma is currently ongoing in Australia.⁹⁹ With the increasing sophistication of interventional radiology and nuclear medicine techniques, there is no doubt that radioembolization to sites other than the liver will continue to expand in the years to come.

Combination therapies with radioembolization

⁹⁰Y microsphere radioembolization is an effective treatment modality for locoregional control of HCC, but the abscopal effect, if any, may not address systemic disease adequately. Various studies have utilized concomitant radioembolization with conventional systemic chemotherapy in order to address this issue.^100–102 An example would be the SIRFLOX study; a randomized controlled trial in which a combination of ⁹⁰Y resin microsphere radioembolization + FOLFOX (oxaliplatin, leucovorin and 5-fluorouracil) chemotherapy regimen was compared against FOLFOX chemotherapy in unresectable liver metastases from metastatic colon cancer. Results from the trial showed mixed results; the trial did not meet the primary end point, as the addition of radioembolization to standard chemotherapy failed to improve overall PFS. However, its secondary end point was met, as it demonstrated that treatment with radioembolization extended the median liver PFS although it has minimal effect on extrahepatic disease. In view of this, more studies with different chemotherapy drugs are needed, as there is still a great potential for combinatorial therapy to be highly efficacious.¹⁰³ With advances in molecular targeted therapy, there are now even more possibilities for systemic therapy in combination with radioembolization. A case study was reported using radioembolization in combination with aflibercept (vascular endothelial growth factor inhibitor), with a folinic acid, flurouracil, irinotecan (FOLFIRI) regimen for the treatment of a metastatic CRC.¹⁰⁴ Another example is a recent multicentre Phase II trial examining the safety and efficacy of radioembolization followed by the multikinase inhibitor sorafenib.⁷¹

CONCLUSION

Radioembolization using ⁹⁰Y microspheres has seen incremental use in the treatment of both primary and secondary hepatic tumours as well as extrahepatic sites. In particular, there is emerging evidence regarding the efficacy of radioembolization for patients with locally advanced HCC and/or PVTT in which a paucity of efficacious treatment modalities currently exist. There have also been continuous advancements in personalized nuclear medicine and interventional radiology techniques for the improvement of radioembolization outcomes. Further Phase II and III clinical trials involving radioembolization either as monotherapy or as combination therapy are ongoing. Given the great potential of radioembolization, there is an impetus to explore additional avenues to enhance its safety and efficacy. This will expand the spectrum of use for radioembolization, opening more doors for those with few treatment options.

FUNDING

Dr Kao has previously received research funding from Sirtex Medical Singapore. Dr Ng has received grants as a co-investigator from Sirtex Medical Singapore. Professor Chow has received honorariums and research grants from Oncosil Pte Ltd, Psivida Pte Ltd, Sirtex Medical and Bayer Pharmaceutical.