Clinical care and technical recommendations for 90yttrium microsphere treatment of liver cancer


  • S-C Wang, MB BS, BSc (Med), FRANZCR, FAMS; L Bester, MB ChB, BSc (Hon) Pharm, M Med Rad (D), MFGP, FRANZCR, FACP; JP Burnes, MB BS, FRACR; JE Clouston, MB BS, FRANZCR; TJ Hugh, MD, FRACS; AF Little, MD, MS, MMed, FRANZCR, FRCR (London); RTA Padbury, MB BS, PhD, FRACS; D Price, MB ChB, FFRad (D)(SA), FRANZCR.

  • Conflicts of interest: See Acknowledgements section.

Professor Shih-chang Wang, Department of Radiology, Westmead Hospital, Hawkesbury Road, Westmead, NSW 2145, Australia.


Selective internal radiation therapy (SIRT) with 90yttrium microspheres is a relatively new clinical modality for treating non-resectable malignant liver tumours. This interventional radiology technique employs percutaneous microcatheterisation of the hepatic arterial vasculature to selectively deliver radioembolic microspheres into neoplastic tissue. SIRT results in measurable tumour responses or delayed disease progression in the majority of eligible patients with hepatocellular carcinoma or hepatic metastases arising from colorectal cancer. It has also been successfully used as palliative therapy for non-colorectal malignancies metastatic to the liver. Although most adverse events are mild and transient, SIRT also carries some risks for serious and – rarely – fatal outcomes. In particular, entry of microspheres into non-target vessels may result in radiation-induced tissue damage, such as severe gastric ulceration or radiation cholecystitis. Radiation-induced liver disease poses another significant risk. By careful case selection, considered dose calculation and meticulous angiographic technique, it is possible to minimise the incidence of such complications to less than 10% of all treatments. As the number of physicians employing SIRT expands, there is an increasing need to consolidate clinical experience and expertise to optimise patient outcomes. Authored by a panel of clinicians experienced in treating liver tumours via SIRT, this paper collates experience in vessel mapping, embolisation, dosimetry, microsphere delivery and minimisation of non-target delivery. In addition to these clinical recommendations, the authors propose institutional criteria for introducing SIRT at new centres and for incorporating the technique into multidisciplinary care plans for patients with hepatic neoplasms.


Primary and secondary liver neoplasms represent a considerable proportion of solid tumours managed by multidisciplinary oncology teams. However, the significant functional reserve and unique vascular arrangement of the liver have facilitated the development of novel, targeted therapies for hepatic tumours. A recent addition to this therapeutic armamentarium is selective internal radiation therapy (SIRT), or radioembolisation with 90yttrium-bonded microspheres. These microspheres may be composed of resin (SIR-Spheres®, Sirtex Medical, Lane Cove, Australia), or glass (TheraSphere®, MDS Nordion, Ottawa, Canada). However, outside of North America, only SIR-Spheres are widely available. Although our experience and discussion in this paper deal primarily with resin microspheres, the general principles enumerated in this document are relevant to any radioembolisation procedure.

SIR-Spheres are biocompatible resin microspheres of 20–60 µm diameter, loaded with the pure beta-emitting radioisotope, 90yttrium.1,2 These microspheres are selectively infused into the hepatic arterial circulation via percutaneous microcatheterisation. As liver tumours derive 80–100% of their blood flow from the hepatic arterial system, hypervascular neoplastic tissue preferentially takes up microspheres via lobar or regional arterial branches.3 Imaging workup therefore necessitates detailed visualisation of the entire liver vasculature and flow patterns, including characterisation of anatomic variants and detection or deduction of collateral flows into extrahepatic tissues.4–7 Administration may require selective occlusion of non-target vessels and attention to regional flow dynamics during microsphere delivery.7 The infused microspheres lodge permanently in targeted vascular beds, simultaneously reducing bloodflow while delivering 90Y brachytherapy.2

Radioembolisation has been successfully employed in the treatment of both hepatocellular carcinomas (HCC) and metastases secondary to colorectal cancer (mCRC), while investigations continue in other indications such as metastases arising from neuroendocrine tumours.1,8 A recent meta-analysis of 90Y radioembolisation trials found that a median 85% of mCRC and HCC patients treated with resin microspheres exhibited ‘any response’ (complete response, partial response or stable disease).9 Although non-liver adverse events tend to be infrequent and mild, SIRT also entails the potential for adverse outcomes which may be severe and, on occasion, fatal.2,10 In particular, owing to the complexity of the hepatic vasculature and its anatomical connections, the procedure carries an inherent risk that microspheres may lodge in extrahepatic vascular beds such as the stomach, duodenum, gallbladder or pancreas.11,12 The resultant irradiation may produce inflammation, ulceration, or frank necrosis of these structures.13 Intratumoural or intrahepatic arteriovenous shunting may lead to excessive delivery of microparticles to the lungs, potentially inducing fatal acute pulmonary oedema or clinically significant pulmonary fibrosis.14

Of particular concern, excessive exposure of uninvolved hepatic parenchyma to radiation may cause cumulative, irreversible liver damage.11,12,15 However, this phenomenon has also been deliberately exploited for therapeutic purposes. Radiation-induced shrinkage of over-treated liver can be employed both to treat localised hepatic neoplasia and to allow sufficient time for compensatory hypertrophy to develop in the untreated part of the liver. In turn, this may enable definitive partial hepatectomy to be performed in about 20% of otherwise surgically untreatable patients. This approach has been termed ‘radiation lobectomy’, and is conceptually analogous to portal vein embolisation, but can be more effective in producing contralateral hypertrophy, especially in cirrhotic livers.16

The potential for serious complications of radioembolisation mandates several requirements for the safe delivery of 90Y microspheres. In addition to adequate training and ongoing clinical diligence on behalf of individual practitioners, the conduct of SIRT also places demands on the equipment, personnel and protocols in place at each treating facility. As approximately 250 patients per year undergo SIRT at a growing number of Australian centres (Sirtex Medical, data on file), it is pertinent to formalise both the procedural and institutional aspects of radioembolisation in order to optimise patient outcomes and safety. To this end, a panel of experienced interventional radiologists and oncology surgeons convened to establish best-practice criteria for the conduct of SIRT. We hope that the following recommendations will serve as a reference for all clinicians and institutions utilising – or contemplating –90Y microsphere radioembolisation for the treatment of patients with liver cancer.

Multidisciplinary teams and the continuum of care

SIRT should ideally be undertaken at centres that employ a multidisciplinary approach to planning, delivering and reviewing cancer treatment, or upon referral from a multidisciplinary team familiar with the procedure (see Table 1 for suggested team composition).17 In particular, multidisciplinary team members should be consulted with regard to the likely interactions between SIRT and any prior, concurrent or planned biological, chemotherapeutic, locoregional ablative, surgical or external beam radiation therapies.7,12,17 All candidates for 90Y radioembolisation should preferably be discussed by a multidisciplinary team – regardless of the referring clinician or centre – and should remain under team review both during and after treatment. As radioembolisation is generally only one aspect of the continuum of care for patients with liver tumours,17 primary responsibility for managing each case should be discussed at the stage of referral. While the interventional radiologist will play a clinical role in assessment and delivery of radioembolisation, ongoing care will generally return to the referring clinician or another nominated member of the multidisciplinary team.

Table 1.  Mandatory and desirable facilities and personnel for sites undertaking 90Y radioembolisation17,18,19
  1. 99mTc MAA, 99mtechnetium macro-aggregated albumin; DSA, digital subtraction angiography.

 Multidisciplinary team approach to reviewing liver cancer patients, including the interventional radiologist and at least three of:
 • hepatic surgeon
 • medical oncologist
 • radiation oncologist
 • nuclear medicine physician
 • pain physician or anaesthetist
 • gastroenterologist/hepatologist
 • medical physicist/radiation safety officer
 Dedicated oncology nursing staff
 Established radiation safety, spill, contamination and disposal protocols
 On-site triple-phase CT, DSA and SPECT 99mTc MAA equipment
 Admitting and clinic rights for interventional radiologists
 On-site or consultant medical physicist
 On-site MRI

The complexity and potential complications of SIRT make it important for the interventional radiologist to consult with each patient both before and after the procedure.20 Referral for SIRT should never imply an obligation to treat the patient. Once a candidate is referred for 90Y radioembolisation, the interventional radiologist should undertake a clinical appointment with the patient to explain the possible outcomes and risks of the workup and procedure, and to assess the patient's suitability and fitness for radioembolisation.20 The interventional radiologist should also undertake a post-procedural assessment of each patient and – regardless of clinical outcome – discuss all cases at multidisciplinary team reviews in order to better inform subsequent treatment choices and procedures.

This evolving clinical role for interventional radiologists will require revisions to admitting rights, outpatient clinic access, procedure time and review appointments. The authors acknowledge that these practices will entail changing attitudes from radiologists themselves, other members of the cancer team and hospital administrators.

Appropriate facilities

The assessment, planning, treatment and follow-up of patients undergoing SIRT predicates specific requirements for each treating institution (Table 1). The first prerequisite is to recognise that radioembolisation is not an isolated procedure. Rather, it entails interaction across the centre – including diagnostic imaging, nuclear medicine, radiation safety, the multidisciplinary care team, oncology nursing staff, discharge and follow-up procedures, and hospital administration. Even if all of these resources are not accommodated at one site, their services must be coordinated within a consistent, accountable, patient-centred system. The process of negotiating the clinical place of 90Y microsphere therapy and integrating its requirements into wider institutional protocols must be undertaken prior to treating the first patient.

Radioembolisation itself is a complex and technically demanding procedure, and emerging centres commonly underestimate the expertise required to safely treat patients. In addition to the mandatory training provided by the sponsor, it is strongly recommended that all interventional radiologists who contemplate adopting SIRT should consult a clinical mentor experienced in the technique.18,21 The initial instruction should ideally take place at centres of excellence compliant with guidelines prepared by appropriate professional bodies, such as the Interventional Radiology Society of Australasia or the Australian and New Zealand Hepatic, Pancreatic and Biliary Association. Once ready to deliver microspheres at their own facility, the inaugural user should conduct a series of procedures both to facilitate their own ‘learning curve’ and to adapt local resources and protocols prior to broadening access for additional operators.20 In addition, problematic cases can be reviewed through electronic case-related consultation by more experienced mentors, even months after the initial training is conducted. Mentors for such training should be willing to respond to subsequent consultation requests in a timely fashion.

Patient selection and diagnostic imaging workup

Candidates for SIRT include patients with one or more unresectable liver tumours that do not respond adequately to other treatment modalities, such as transplantation, radiofrequency ablation or transcatheter arterial chemoembolisation. Patients known or anticipated to have poor tumour response to chemotherapy may also be considered for radioembolisation.22 However, patients must demonstrate sufficient functional liver reserve and minimal extrahepatic metastases prior to commenting SIRT, and should be anticipated to survive for at least 3 months after the procedure. Additional patient history, laboratory tests and clinical factors to be considered by the interventional radiologist are iterated in Table 2.

Table 2.  Suggested patient selection and exclusion criteria for 90Y radioembolisation2,17,20,22
  1. ALT, alanine amino transferase; AST, aspartate amino transferase; ECOG PS, Eastern Co-operative Oncology Group performance status; HCC, hepatocellular carcinomas; SIRT, selective internal radiation therapy.

Acceptable presentations and history include
 Life expectancy >3 months
 Liver-only or liver-dominant disease
 Partial, lobar invasion of the portal vein
 Tumour burden <50% of liver volume
 Tumour(s) not amenable to thermal ablation or surgical resection
 Child–Pugh status of A, or at most early B in cirrhotic patients
 Serum ALT or AST ≤5 times upper limit of normal
 Caution if prothrombin time extended
 Serum bilirubin ≤35 µmol/L
 Serum albumin ≥30 g/L
 Minimal comorbidities
Acceptable HCC parameters
 Serum alpha-fetoprotein >400 µg/L or imaging diagnostic of HCC
 History of viral or alcoholic cirrhosis
 Prior radiofrequency ablation
 Prior liver segmentectomy
 Prior chemoembolisation
Acceptable mCRC parameters
 Serum carcinoembryonic antigen >2.5 µg/L
 Prior standard-of-care systemic chemotherapy
 ECOG PS 0–2 or Karnofsky index ≥60%
Contraindications to SIRT
 Ascites or clinical liver failure
 Child–Pugh status late B or C
 Prior external beam radiotherapy to the liver
 Prior extensive liver resection or any bilio-enteric anastomosis
 Concurrent or prior capecitabine chemotherapy (within preceding 2 months)
 Obstructed bile duct or extensive portal vein thrombosis, or portal or biliary stent in situ
 Anticipated lung exposure to 90Y radiation >30 Gy based on pre-procedural 99mTc MAA scan
 Anticipated reflux of microspheres into arteries supplying the stomach, pancreas or duodenum, based on pre-procedural angiography

Imaging technology and expertise vary between institutions, but a minimum diagnostic workup for SIRT candidates involves a recent (within 4–6 weeks of angiographic workup) multi-phase CT of the liver, digital subtraction angiography (DSA) and gamma camera 99mtechnetium macro-aggregated albumin (99mTc MAA) scanning. Where available, additional techniques of value include CT hepatic angiography with intra-arterial catheter contrast delivery, CT of the diffusion-weighted magnetic resonance imaging (MRI) liver scans, single-photon emission computed tomography (SPECT) and positron emission tomography (PET).5,17,20

The liver vasculature and its connections with the gastrointestinal bed are both anatomically complex and highly variable between individuals.4,6,7 Furthermore, prior hepatic resection or multiple episodes of chemoembolisation frequently cause significant changes in collateral vascular supply that must be taken into consideration.23 In assessing each candidate for SIRT, the treating interventional radiologist should meticulously document both the anatomical distribution of the hepatic arterial circulation and its associated flow dynamics. Thorough characterisation of these vessels and their accessory branches may be laborious, but is essential to maximise the targeted delivery of radioembolic microspheres to tumours while minimising deposition in non-target tissues.7,18 The treating interventional radiologist should consider seeking opinions from skilled colleagues if there is any uncertainty as to which vessels should be embolised, and also when interpreting any ambiguous post-embolisation films.

In addition to the number, location and volume of tumours, pertinent features to assess include the total volume of the liver, portal vein patency, relative hypervascularity of the neoplastic regions and any extrahepatic lesions. Multiplanar reconstructions of CT with arterial contrast are very useful in identifying non-target vessels and establishing the patency of the portal vein.7 A recommended protocol is multislice CT reprocessed to 0.5 or 1.0 mm in order to permit multiplanar reconstructions. The radiologist should pay particular attention to the degree of tumour enhancement relative to the amount of contrast delivered and the number of phases performed, as weakly enhancing (i.e. relatively hypovascular) lesions may not allow for a suitable degree of preferential uptake of microspheres from the hepatic arterial circulation.5 Successful treatment of refractory or non-enhancing metastases has been documented after segmental or lobar administration of microspheres, but there are insufficient data to safely proceed with SIRT if multiple non-enhancing lesions are present.24 Furthermore, precise estimation of total and segmental liver volumes – plus the volume of all tumours – is necessary for accurate dosimetry calculations, especially if partition model dosimetry is used.7,25

Angiography and non-target vessel embolisation

SIRT is a flow-directed therapy that relies upon neoangiogenesis and increased blood flow in malignant tumours to preferentially take up 90Y microspheres from the hepatic arterial circulation. This principle underlies the tumour and vessel mapping requirements for successful radioembolisation, requiring both selective embolisation of non-target vessels (where indicated) and carefully positioned catheter tip placement during infusion.7

Owing to its superior resolution of small vessels, catheter angiography remains the primary medium for characterising the hepatic and mesenteric vasculature.7 Meticulous technique is required to map all relevant vessels in the hepatic, gastric and mesenteric beds, including anatomical variants, extremely small branches and collateral vessels (Table 3).4,6,7 Whether non-target vessels are occluded during the mapping procedure, or on the day of 90Y microsphere infusion, is at the discretion of the interventional radiologist. Embolisation prior to the day of SIRT is generally preferred, as it allows time for adequate assessment of altered flow patterns and dosimetry, thus reducing the risk of last-minute cancellation and associated waste of 90Y microspheres.12 However, because selective occlusion may result in adaptive flow patterns through non-embolised collateral vasculature, a maximum of 2–3 weeks should separate vessel mapping and administration of microspheres.17

Table 3.  Minimal arterial catheterisation paths and vessels of interest during angiography prior to radioembolisation4,6,7
Superior mesenteric arteryGastroduodenal arteryLeft hepatic arteryRight hepatic arteryOther vessels
Parasitised flow to the liverParasitised flow to the liverLeft inferior phrenic arteryMiddle hepatic arteryAdrenal artery
Replaced right hepatic artery(Accessory) cystic arteryAccessory left gastric arterySupraduodenal arteryPhrenic arteries
Accessory right hepatic arteryAccessory hepatic arteriesInferior oesophageal arteryFalciform arteryLumbar arteries
Replaced proper hepatic arterySuperior pancreatoduodenal arteryRight gastric arteryCystic arteryColic or omental branches
Replaced common hepatic artery Falciform artery Renal or renal capsular artery
Coeliacomesenteric trunk   Inferior mammary artery
Portal vein patency   Accessory vessels

The aim of embolising non-target vessels is to minimise the flow of 90Y microspheres into both uninvolved liver parenchyma and extrahepatic tissues. In order to achieve this goal, however, it is not necessary to maximally occlude all non-target vessels. Indeed, this strategy may result in recruitment of collateral vasculature with unknown connections, particularly vessels arising from the gastroduodenal or right gastric arteries. Such redistribution phenomena may result in unpredictable deposition of microspheres with undesirable clinical consequences. For these reasons, selective or superselective embolisation is the preferred approach, tailored to the vascular pattern and tumour burden of each individual patient. The choice of embolic material depends on the size of the vessel to be occluded, the timeframe for the occlusion and the nature of the distal vascular tree.

In determining which vessels to occlude, the embolic load of the microspheres themselves should be taken into account, as occlusion during radioembolisation may be greater than anticipated. Deposition of microspheres into the right gastric and gastroduodenal arteries must be avoided, either by selective embolisation, or by ensuring that microspheres are infused distal to these vessels while scrupulously avoiding reflux.4,6,7,17,21 Particular attention should be paid to the gastroduodenal artery, as its accessory hepatic vessels may communicate with the tumour, while the flow outcome from embolising this vessel can be difficult to predict.4 The cystic artery should not routinely be embolised, but this may be indicated in some cases, for instance if it is small or if its origin is distal to the anticipated site for microsphere infusion into the right hepatic artery.4,6,11

Other vessels to consider for prophylactic embolisation include the oesophageal, falciform and accessory phrenic arteries, in addition to anatomical variants of concern such as the supraduodenal or retroduodenal arteries.4,6,7,20 Prior repeated chemoembolisation in refractory or recurrent hepatocellular carcinoma may lead to parasitic neovascularity arising from vessels, including the inferior phrenic artery; an omental branch or branch of the superior mesenteric artery; internal or inferior mammary arteries; renal or renal capsular arteries; or adrenal, intercostal, cystic, gastric or lumbar arteries.23 Careful mapping of these unfamiliar types of neovascularisation is required to determine whether SIRT can be safely delivered in such patients.

Following embolisation, a CT hepatic angiogram may be of value in determining the selective arterial supply to the tumour and any residual flow to non-target vessels. All patients, however, must undergo a 99mTc MAA scan after vessel occlusion to exclude unsuitable candidates for radioembolisation.18,26,27 As MAA particles are of similar size to 90Y microspheres, their selective uptake allows the interventional radiologist to calculate the proportional arterial flow into the tumour versus normal tissue (T:N ratio).12,26–28 Depending upon how the MAA is delivered, flow patterns can also be predictive for extrahepatic deposition into the gallbladder, pancreas, stomach or duodenum, plus any arteriovenous shunting to the lungs during radioembolisation.20,26 MAA should ideally be delivered selectively into the hepatic circulation to simulate as closely as possible the planned 90Y radioembolisation, allowing accurate determination of any likely pulmonary breakthrough. Although MAA is biodegradable, the mass of particles injected may temporarily alter flow dynamics. Breakthrough scans should therefore generally not take place on the same day as 90Y radioembolisation.12

90Y microsphere dosimetry

The principles and practice of dosimetry for SIRT are complex and can only be summarised here. The underlying premise for 90Y microsphere dosimetry is to ensure that the liver parenchyma exposure does not exceed 70 Gy, while at least 120 Gy must accumulate within neoplasms to deliver a dose-dependent tumouricidal effect.29–31 These bulk tissue thresholds are derived from external beam radiotherapy experience, where cumulative doses exceeding 70 Gy to the liver typically cause irreversible radiation injury. More recently, it has been suggested that cumulative exposure over two or more SIRT treatment cycles may safely – and indeed greatly – exceed 70 Gy in carefully selected cases, especially in HCC.32

Estimation of the radiation dose actually delivered to neoplastic tissue is, however, problematic. The tumour type and the heterogeneity of vasculature within each lesion will affect dosing: HCC are generally more hypervascular than mCRC and often receive a much higher dose of radiation than anticipated from simple body surface area (BSA) dosimetry estimates.27 Furthermore, 90Y microspheres usually distribute relatively homogeneously throughout HCC, whereas they tend to congregate on the more vascular perimeter of mCRC lesions.28,33 There is also considerable variability at the micro level, with clustering of spheres inside neoplastic tissue producing localised exposures of up to 3000 Gy within the average radiation range of 2.5 mm.28,33,34 This shallow tissue penetration, however, means that accurate external measurement of localised beta radiation is problematic once the microspheres are infused.26

Empirical dosing of 90Y microspheres is now actively discouraged. The manufacturer's package insert suggests two alternative methods to calculate appropriate dosing: BSA or partition modelling.20,25 There are no data to compare the clinical effectiveness of these methods, but the BSA approach was employed in the trials upon which regulatory approval was granted and therefore must remain the recommended formula for calculating dose in relatively straightforward cases.8,17 BSA dosimetry can be utilised both for whole liver and lobar/segmental treatment,10 but the relative insensitivity of this method suggests that patients whose lung shunting fraction exceeds 20%, whose serum albumin is lower than 25 g/L or whose serum total bilirubin exceeds 35 µmol/L should not be treated with radioembolisation. The partition model is more sensitive to regional volumes, but requires that the tumour should be localised within a discrete area that is amenable to clear SPECT visualisation.19,25,30,35 In practice, these limitations tend to see the partition model employed for HCC dosimetry, while BSA is largely utilised for mCRC.19 Partition model dosimetry is quite time consuming, and many clinicians who employ this method compare it with a simpler BSA calculation for the same patient in order to determine the optimal infusion volume.

Administration of 90Y microspheres

Owing to the diversity of presentations between patients and variability in the training of interventional radiologists, it is impossible to prescribe a single correct procedure for administering 90Y microspheres. However, there are important conventions that should be followed and several practices that the authors actively discourage. In particular, both local and sponsor protocols for radiation containment and safety should be stringently adhered to.18,19

Both accurate dosimetry and a pre-defined strategy to avoid non-hepatic deposition of microspheres are essential preconditions for SIRT. Additionally, it must always be borne in mind that arterial flow dynamics vary considerably throughout the radioembolisation procedure.7 This is due both to the embolic load of the microspheres, and to vessel reactivity in the presence of the microcatheter and the perfusion pressure applied during infusion.7 Large hypervascular HCC has a substantial ‘sump’ effect, and it is unusual to encounter any alteration in flow during infusion of such tumours, even with very high doses (for instance, exceeding 3 GBq). Conversely, low-volume mCRC lesions have a very small incremental hypervascular capillary capacitance relative to the surrounding normal liver. It is therefore not uncommon to saturate the vascular bed before the entire calculated dose can be delivered.

The administration device itself contributes to this variability as its air-buffered system allows no control over the pressure at the catheter tip, while suspension of the spheres – and thus their distribution within the infusion fluid – also cannot be fully controlled once the procedure has commenced.20 These limitations of the apparatus, however, only increase the responsibility on the clinician to ensure that administration is as smooth as possible.20 Slow injection and constant vigilance are essential to minimising any turbulent flow or unanticipated jets, which may dislodge the catheter tip from its desired location and/or propel microspheres into unintended vessels.20 It is crucial to sustain sufficient pressure and flow to maintain a fairly uniform microsphere suspension and to prevent silting or ‘settling out’, which can occur in the vial, delivery tubing or catheter. Most users currently employ a 10- or 20-cm3 syringe to administer tiny ‘pulsed injections’ of 0.1–0.2 mL, but more recently, it has been suggested that smooth, slow, non-pulsed injection via a 2-cm3 syringe can maintain adequate particle suspension without producing any turbulent jets of microspheres at the catheter tip (YH Kim, Korea Medical University, personal communication).

In order to maintain fluoroscopic surveillance of changing flow dynamics throughout the procedure, it is suggested that microspheres should be administered in multiple brief infusions sandwiched between boluses of contrast medium.18,20 According to the manufacturer's instructions, contrast medium and microsphere infusion fluid should never be mixed, so the line must be flushed through with sterile water between administrations of each agent.19

Infusion of the microspheres into one or both lobes can generally be accomplished in a single procedure unless the patient has impaired hepatic function, in which case two procedures separated by 30–45 days may be required.17 Unilobar or sequential bilobar administration of microspheres is recommended over whole-liver infusion, as this reduces peri- and post-procedural complication rates.12,17,20 The preferred infusion points are the left and right hepatic arteries or their distal segments. Administration into the right hepatic artery should only occur, however, if the target branch is at least 1 cm distal from the origin of a non-occluded cystic artery. Microspheres should not be administered into the common hepatic artery as its multiple collateral vessels allow little control over flow into uninvolved parenchyma or extrahepatic tissues.12 Likewise, administration into the proper hepatic artery is generally not recommended, but may be feasible if its origin is sufficiently distal from the junction with the common hepatic artery to permit a safe degree of antegrade flow.12

A common approach to bilobar administration is to catheterise the most difficult vessel first – generally the left hepatic artery. After passing the microcatheter into the artery, the guide wire can be retracted and the spheres infused. After flushing through the system with sterile water, the catheter can be retracted and then advanced into the right hepatic artery if possible. If this is not feasible, the guide wire can be re-threaded through the microcatheter until correct placement is obtained. The now-radioactive guide wire can now be removed and discarded; further spheres can then be infused prior to withdrawal of the radioactive microcatheter for safe disposal.

Prophylactic anti-emetics can be administered on the day of radioembolisation and intravenous pain relief should be on hand throughout the procedure.19 In cases where the cystic artery is embolised, ischaemic cholecystitis refractory to treatment may eventuate, occasioning urgent cholecystectomy. Although there is no published evidence to support its clinical benefit, it is the consensus opinion of the authors that prophylactic antibiotics should be administered whenever the cystic artery is embolised.6,11 It is also suggested that an 11-min SPECT or planar scintigraphy Bremsstrahlung scan be conducted within 30 h of the 90Y microsphere infusion to ascertain any non-target deposition and guide subsequent follow-up.21,26

Patients should be monitored and treated for a post-embolisation syndrome that may occur within 1–14 days after SIRT, including abdominal pain or flu-like symptoms, such as fatigue, nausea, fever or chills. On occasion, this syndrome is sufficiently severe to warrant admission and parenteral therapy, including antibiotics, analgesics, anti-emetics and/or oral corticosteroids where indicated.10,12,19,20 Owing to the possibility of delayed adverse effects of treatment, patients should be informed of the potential for such events and urged to contact their oncology team if they experience malaise, discomfort or other radiation sequelae.

Minimising risk of gastric or bowel ulceration

Unrecognised or inadequately occluded hepaticoenteric communication may result in the dispersal of microspheres into the gastrointestinal vasculature, leading to irradiation of the gastric or duodenal mucosa.21 Both the anatomical location of gastric lesions and the low tissue penetration of beta particles confirm that such irradiation arises from microspheres lodged within the vasculature of the stomach, rather than from adjacent liver or tumour tissue. This contention is supported by published biopsy reports, which identify distinctive dark staining around resin microspheres within radiation-induced gastric or duodenal ulcers.36 Although 94% of the 90Y dose will have decayed within 11 days of the procedure,2 radiation-induced mucosal lesions may take 2–6 weeks to manifest. Symptomatic ulceration arising from this unintended exposure can be severe – sometimes with potentially life-threatening arterial bleeding – and is often resistant to pharmacological management.20

Prophylactic treatment to minimise the severity of gastric ulceration is therefore an essential component of SIRT, but it does not obviate the requirement for scrupulous vessel mapping, occlusion and administration of microspheres.13 Although there is no evidence for ulcer prophylaxis in the setting of SIRT, extrapolation from gastric ulcer studies suggests that routine proton-pump inhibitor therapy may be valuable.19,37,38 Omeprazole has been commonly used in this indication, but its efficacy may be compromised by polymorphisms in cytochrome P450 enzymes 2C19 and 3A4 in up to 20% of patients, particularly those of Asian ethnicity.38 Most centres administer a high-dose omeprazole regimen to overcome this problem (20 mg twice daily for 2–4 weeks). However, as rabeprazole is only minimally metabolised by CYP2C19 and 3A4, an alternative approach may comprise high-dose rabeprazole sodium (20 mg once daily) from 1 week prior to SIRT until 6 weeks post-procedure.19,38 Clinicians may also wish to consult with their local gastroenterologist regarding the value of prophylactic Helicobacter pylori eradication as a means to further reduce risk of ulceration.37

If an immediate post-radioembolisation Bremsstrahlung SPECT scan shows complete confinement of the infused radioactive microspheres to the liver, proton pump inhibitor therapy may not be necessary. However, this scanning procedure is neither standard nor widely available, and in general, routine administration of omeprazole or rabeprazole is strongly recommended. As proton pump inhibitor therapy may mask symptoms of an incipient ulcer, the possibility of emergency gastric complications should be borne in mind by the clinician responsible for the patient's long-term care. Gastroscopy is not routinely undertaken in patients who have undergone SIRT, but may be appropriate even at a low index of suspicion.20,37

Monitoring for radiation-induced liver damage

Many patients with HCC or heavily pretreated mCRC present with abnormal liver function tests. In addition, a rapid elevation in hepatic enzymes – and sometimes bilirubin – is observed almost universally in the first 6–12 weeks following SIRT, and should not be a cause for undue alarm.10 However, approximately 4% of patients develop radiation-induced liver damage within 4–8 weeks of radioembolisation, although this may not become apparent for several months after the procedure.10,15 Individuals at risk for radiation hepatotoxicity following SIRT include those with compromised baseline hepatic function – especially elevated serum bilirubin or frank cirrhosis – and those with reduced liver volume following surgical resection. Patients who receive higher-than anticipated doses of microspheres, particularly following prior chemotherapy, are also at increased risk.10,11,15 Additional factors, including recent capecitabine administration, substantial previous chemotherapy or extensive liver involvement, may exacerbate hepatic susceptibility to radiation.10,15

Presentations of radiation-induced liver damage vary, but include hepatomegaly, jaundice and derangement of liver function tests, leading subsequently to anicteric ascites, organ shrinkage, significant and permanent decline in liver function, or fulminant hepatic failure.10,15,21,31 Therefore, all clinicians responsible for the care of a patient with liver cancer following radioembolisation must remain alert to the possibility of iatrogenic hepatotoxicity, which may take 3–6 months to manifest. Liver function tests should be first assessed within 4–8 weeks post-procedure and followed regularly.15 If radiation-induced liver damage is suspected, treatment options including diuretics or corticosteroids should be discussed by the multidisciplinary team.12,15,19,21 In rare cases of severe toxicity refractory to treatment, a transjugular intrahepatic portosystemic shunt may be considered.

Patient follow-up

Responsibility for the long-term follow-up of patients – including an awareness of specific risks and adverse events associated with SIRT – must be agreed at the stage of initial referral and clearly communicated both to the responsible clinician and the patient. Whichever clinician takes on this duty of care, the patient should be reviewed by the treating interventional radiologist at least once following radioembolisation. This appointment should occur within 6–12 weeks of SIRT and should include a review of both imaging and clinical parameters.12,20

The minimum data for review include relevant tumour markers plus CT or PET/CT scans to assess the oncological response, in addition to clinical chemistry and liver function tests to evaluate any potential for radiation-induced liver damage.12 Follow-up scans are best undertaken either at the treating centre or within its referral network, in order to ensure that interpretation of images is consistent with the mode of action of SIRT.18 If these scans are performed more remotely, patients can be asked to attend specific high-quality CT centres conversant with the follow-up imaging features of SIRT. Ideally, these remote scans should be additionally reviewed by the treating team to permit detailed comparison with pre-treatment imaging.

CT scans performed less than 3 months post-procedure tend to show persistent tumour perfusion and care should thus be taken in their assessment. Typically, 3-month post-treatment CT scans display decreased attenuation; as this phenomenon is apparently reversible, it should be clearly distinguished from indicators of tumour recurrence when interpreting follow-up imaging.21,39 In assessing tumour response, reductions in tumour markers, lowered metabolic activity on 18fluorodeoxyglucose-PET/CT or an increase in the necrotic centre of the neoplasm are more appropriate indicators than measurement of linear dimensions, which may remain static or even increase if tumour necrosis leads to peritumoural oedema or haemorrhage.12,26,39,40 Patients should undergo additional scans at 6 and 12 months under the care of the primary clinician. Institutions performing SIRT should ensure that all cases are periodically reviewed at multidisciplinary team meetings attended by the interventional radiologist.

Conclusions and future directions

SIRT with 90Y microspheres offers clinically meaningful benefits for selected patients with liver cancer. The nature of the procedure and the possibility of both intra- and extra-hepatic serious adverse events, however, means that the technique should only be offered by skilled practitioners working with a multidisciplinary team at an appropriately equipped facility.

In addition to growing individual and institutional clinical experience, it is hoped that a national or international registry of patients undergoing SIRT can be established to further inform patient selection and management.17 Furthermore, studies to compare dosimetry methods, to validate or improve current approaches to accurate dosimetry, and to more closely correlate delivered doses with clinical outcomes would be welcomed. We hope that both existing and emerging centres will adopt the recommendations laid out in this article as a means to standardise protocols, minimise risks and optimise outcomes for patients with inoperable liver cancer.


The content for this paper was discussed and agreed by the authors at an advisory board meeting supported by Sirtex Medical. A freelance medical writer, Peter Hobbins, undertook the writing of the first draft and coordinated author amendments. Each author contributed to reviewing and revising the manuscript until group consensus on the content was achieved. While Sirtex Medical was invited to review the paper prior to submission to ensure accuracy of technical and regulatory information, the authors take full responsibility for the content and expression of the final manuscript. None of the authors is a shareholder in Sirtex Medical. Shih-chang Wang and Lourens Bester have acted as advisors and consultants to Sirtex Medical, and have been proctors and mentors for emerging treatment centres in Asia. Andrew Little and John Clouston have acted as advisors to Sirtex Medical and were clinical participants in the SIRFLOX trial. James Burnes was a co-investigator in the SIRFLOX trial but received no support from Sirtex Medical for this or any other role, except for participation in the meeting at which this paper was created. Apart from attending the advisory board associated with this paper, Thomas Hugh, Robert Padbury and David Price have no affiliation with Sirtex Medical.